With increasing computer-based work burden, prolonged poor sitting posture can result in health issues such as scoliosis. However, current sitting posture recognition systems often require the purchase of additional hardware. The camera-based system can compromise user privacy and be affected by varying lighting conditions. In this paper, we propose a solution to realize a sitting posture recognition system derived from acoustic signals generated by smartphones. Firstly, acoustic signals corresponding to various sitting postures are acquired via the built-in speaker and microphone of the smartphone. Subsequently, an innovative signal segmentation technique based on the adaptive threshold is designed to extract the signals, followed by the creation of a deep learning model for posture recognition. To meet the demands of lightweight deployment, a knowledge distillation compression technique is introduced to compress the model while maintaining its accuracy. The experimental results validate that our sitting posture recognition system has good effectiveness and robustness, making it more universal.

Proper sitting posture is essential for health, as maintaining appropriate torso inclination improves comfort and reduces the risk of cervical and spinal disorders. Despite extensive research on posture detection, fine-grained variations in torso inclination have received limited attention. We propose Torso-Track, a non-intrusive, device-free system for monitoring torso inclination. TorsoTrack leverages wireless respiratory signals to capture posture changes, applies signal restoration to correct abnormal segments, and extracts key features for precise angle estimation. A multi-layer feature extraction network models the relationship between wireless signals and torso inclination, emphasizing critical posture features. To enhance adaptability to new users, TorsoTrack integrates a Reptile-PG meta-learning strategy, improving cross-individual prediction accuracy. Evaluations on 25 participants show that TorsoTrack reliably identifies significant features corresponding to different torso angles, achieving an averageR2of 0.9822, RMSE of 1.6733, and MAE of 0.9977. For previously unseen individuals, the system maintains an averageR2of 0.9033, demonstrating its effectiveness in precise torso inclination assessment.

Motion artifacts (MA) significantly compromise the quality of photoplethysmographic (PPG) signals in wearable devices, hindering the accurate extraction of vital sign parameters. To address the limitations of existing methods in cross-user MA detection and the low accuracy of signal reconstruction, this paper proposes FlexiMA, a two-stage co-optimization framework that operates without reference signals. This framework is designed to tackle two key tasks: MA detection and MA removal. In the detection phase, a task-adaptive diffusion model, IMD-XResNet1D, is developed by integrating meta-learning and diffusion mechanisms, enabling efficient modeling of artifact-related features in new users. In the removal phase, an adversarial variational autoencoder (AVAE) is employed to achieve high-fidelity reconstruction of PPG signals. Experiments conducted on multi-user, multi-scenario datasets demonstrate that FlexiMA achieves 99.68% accuracy in MA detection and maintains a cross-user generalization performance of 99.32%. In the MA removal task, it attains 87.85% waveform structure similarity with an interbeat interval (IBI) error of 4.60 ms, indicating reliable signal reconstruction. These results verify the robustness and adaptability of FlexiMA in dynamic and complex environments, highlighting its strong potential for deployment in wearable applications.

Leather is among the most widely traded textile materials worldwide, and detecting surface defects is crucial to ensuring its quality. Conventional manual inspection of leather surfaces is labor-intensive, time-consuming, and prone to human error. Advances in computer vision have facilitated the development of automated leather surface defect detection. However, vision-based methods remain sensitive to lighting conditions, particularly on high-gloss leather surfaces. In this paper, we introduce Le-Radio, an industrial IoT (IIoT) oriented RF-based system for automated leather defect detection. Le-Radio provides a gloss-robust, ubiquitous, and continuous inspection solution, which is crucial for integrating quality control into smart factory pipelines and advancing the digital transformation of the traditional leather industry. To extract informative representations of leather defects from wireless signals, we employ continuous radio snapshots and refined signal features to differentiate defect patterns. Specifically, we design an imaging algorithm that continuously visualizes defects on moving leather samples using a fixed radar. To enhance Le-Radio’s robustness in diverse environments, we train a transferable model that maintains consistent detection performance across different scenarios. We conduct extensive experiments in three rooms using 120 leather samples to evaluate Le-Radio. Experimental results demonstrate that Le-Radio accurately detects diverse defects in high-gloss leather. Beyond leather inspection, this work establishes Le-Radio as a practical IoT edge sensing node for material surface inspection in the IIoT. By integrating RF-based defect scanning capabilities and a robust radar imaging framework into a unified system, this work establishes RF sensing as a viable and novel paradigm for material surface inspection in the IIoT, particularly in scenarios where optical methods are fundamentally limited.

Private data, when published online, may be collected by unauthorized parties to train deep neural networks (DNNs). To protect privacy, defensive noises can be added to original samples to degrade their learnability by DNNs. Recently, unlearnable examples (Huang et al., 2021) are proposed to minimize the training loss such that the model learns almost nothing. However, raw data are often pre-processed before being used for training, which may restore the private information of protected data. In this paper, we reveal the data privacy violation induced by data augmentation, a commonly used data pre-processing technique to improve model generalization capability, which is the first of its kind as far as we are concerned. We demonstrate that data augmentation can significantly raise the accuracy of the model trained on unlearnable examples from 21.3% to 66.1%. To address this issue, we propose a defense framework, dubbed Armor, to protect data privacy from potential breaches of data augmentation. To overcome the difficulty of having no access to the model training process, we design a non-local module-assisted surrogate model that better captures the effect of data augmentation. In addition, we design a surrogate augmentation selection strategy that maximizes distribution alignment between augmented and non-augmented samples, to choose the optimal augmentation strategy for each class. We also use a dynamic step size adjustment algorithm to enhance the defensive noise generation process. Extensive experiments are conducted on 4 datasets and 5 data augmentation methods to verify the performance of Armor. Comparisons with 6 state-of-the-art defense methods have demonstrated that Armor can preserve the unlearnability of protected private data under data augmentation. Armor reduces the test accuracy of the model trained on augmented protected samples by as much as 60% more than baselines. We also show that Armor is robust to adversarial training. We will open-source our codes upon publication.

Vision transformers have achieved impressive performance in various vision-related tasks, but their vulnerability to backdoor attacks is under-explored. A handful of existing works focus on dirty-label attacks with wrongly-labeled poisoned training samples, which may fail if a benign model trainer corrects the labels. In this paper, we proposeMegatron, an evasive clean-label backdoor attack against vision transformers, where the attacker injects the backdoor without manipulating the data-labeling process. To generate an effective trigger, we employ a local surrogate vision transformer to approximate the victim model and customize two attention-based loss terms: latent loss and attention diffusion loss. The latent loss aligns the last attention layer between triggered samples and clean samples of the target label. The attention diffusion loss emphasizes the attention diffusion area that encompasses the trigger. A theoretical analysis is provided to underpin the rationale behind the attention diffusion loss. Extensive experiments on CIFAR-10, GTSRB, CIFAR-100, and Tiny ImageNet demonstrate the effectiveness ofMegatron.Megatroncan achieve attack success rates of over 90% even when the position of the trigger is slightly shifted during testing. Furthermore,Megatronachieves better evasiveness than baselines regarding both human visual inspection and defense strategies (i.e., DBAVT, BAVT, Beatrix, TeCo, and SAGE).

Millimeter-wave (mmWave) radars are indispensable components of safety-critical advanced driver assistance systems, enabling accurate and weather-resilient environmental sensing for autonomous vehicles. Despite the advanced sensing capabilities, mmWave radars remain susceptible to adversarial attacks, where malicious users attempt to distort the sensing results of victim radars, leading to hazardous driving behaviors. While existing anti-spoofing techniques have been proposed to mitigate specific attacks, they may be ineffective against adaptive adversaries. To address this critical vulnerability, we introduce AttackDeceiver, a novel anti-spoofing system using a phase-shifted interleaving waveform. By comparing range and velocity estimates from two independent virtual channels, our system effectively detects and mitigates the effects of spoofing attacks. In addition, we proactively counter adaptive spoofing attacks by inducing attackers to generate false targets with unrealistic velocity fluctuations. A compact prototype of AttackDeceiver is realized using commercial-off-the-shelf radar kits. Experimental results demonstrate the effectiveness of our system, achieving a remarkable false target recall exceeding$97.9\%$and a significant enhancement in signal-to-interference-plus-noise ratio exceeding$\text{13.46}\,\text{dB}$.

Network Operating Systems (NOS) are being widely deployed on edge devices by cloud service providers to perform fast configurations and offer high availability for new network protocols. However, NOS-enabled networks open the door to intruders that can stealthily corrupt less-guarded programmable switches to launch attacks on the entire network. Traditional centralized intrusion detection systems may neglect anomalous events on NOS-equipped switches and fail to detect such attacks. In this paper, we make the first attempt towards intrusion detection for NOS-enabled networks by designingRetriever.Retrieverfeatures a lightweight local anomaly detection module on programmable switches and a central anomaly assessment module on the central server. The local anomaly detection module selectively traces both system and network events on switches, based on which a provenance graph of events is established. Upcoming events unmatched by the provenance graph are aggregated to construct a suspicious subgraph to report to the central server. The central anomaly assessment module extracts semantic representations from reported suspicious subgraphs and computes their anomaly scores. Large-scale experiments show thatRetrievercan achieve high intrusion detection accuracy (nearly 100%) with low overheads.

When it comes to the marriage of graph neural networks (GNNs) and model extraction attacks, the deployment of GNNs within Machine Learning as a Service (MLaaS) through a publicly pay-per-query API has opened up new attack surfaces. Existing defenses either sacrifice prediction accuracy or fail to thwart more advanced attacks. We investigate this dilemma and discover that fortified models with complex and narrow decision regions are difficult to be reproduced. Nevertheless, complex and narrow decision boundaries are prone to violate the subspaces of neighbor classes under the intrinsic coupling property of graph structure. Furthermore, class-wise representative features within the interior of class-wise subspaces endow the attackers with the capability of functionality replication. Here, we propose a novel model extraction defense, dubbed Decision Boundary-aware Counterfactual Learning (DBCL). DBCL proactively launches counterattacks on potential model extraction attacks, from the very beginning of sensitivity measurement that implicitly detect the malicious queries, such that class-wise representative features embodied in the highly sensitive query batch trigger the demand of worsening their query results unconsciously. Moreover, DBCL draws inspiration from counterfactual learning, aiming at finding the decision boundary-aware adversarial topology perturbations for ambiguously classified query samples, i.e., hard samples, to cross the decision boundary exactly, which introduces the tractive behaviour w.r.t. the inter-connected sensitive samples for class-wise ambiguous topology features. From the graph-structured actionable insights, DBCL innovatively finds the minimum perturbation sufficient for counterfactual learning, without jeopardizing the victim model’s predictive capacity by including confidently classified query samples, i.e., easy samples including sensitive and non-sensitive samples, into their correct classes. Empirically, DBCL shows its effectiveness in reducing the extraction accuracy of the SOTA model extraction attempt with different GNN backbone encoders in evaluating node classification performance. Moreover, we show that DBCL is robust to adaptive model extraction attacks.

In the domain of smart devices, biometric identity authentication has become a leading and crucial technology, mainly because of its improved security and user convenience. Traditional methods frequently depend on complex activities, facial recognition, or passwords, which can be cumbersome and error-prone. This paper presents GrasOpen, an innovative biometric authentication system tailored for door-opening scenarios, utilizing the natural reach-and-grasp motion linked to door handles. The system employs smartwatches with accelerometers and gyroscopes to track and analyze arm movements, ensuring a seamless and intuitive user experience. GrasOpen tackles key challenges by leveraging the unique characteristics of the reach-and-grasp motion, eliminating the necessity for users to remember complex motions, and avoiding redundant actions like waiting for facial recognition. Specifically, GrasOpen initially proposes a lightweight model to discern door-opening actions from daily activities. Then, to realize the robust authentication system, GrasOpen integrates a complementary filter (CF) method to capture diversity-dependent features in door-opening actions. These features are subsequently processed through a modified ConvBoost model for precise user authentication. Experimental results reveal an impressive accuracy of 98.78% for activity recognition and 98.65% for identity authentication, highlighting GrasOpen’s excellence in function and performance. The system’s security and robustness are validated across diverse authentication environments.

Unmanned Aerial Vehicles (UAVs) have emerged as relay nodes to alleviate multi-user traffic congestion in wireless communication networks due to their flexibility, fast deployment and line-of-sight links. However, the selfish attribute of both users and UAVs presents significant challenges to the efficient participation of UAVs in multiple relay tasks, particularly with conflicting profit objectives. In this article, we propose a double auction-based incentive mechanism that incentivizes UAVs to engage in relay tasks while achieving a strategic trade-off between network throughput and market fairness. Specifically, UAVs (sellers) provide relay services to ground users (buyers), with base station acting as an auctioneer responsible for matching user and UAV, charging fees from users, and giving payments to UAVs. The presence of heterogeneous users and UAVs complicates the relay tasks processing. Additionally, designing an effective payment strategy that meets economic properties further exacerbates the difficulty. To address these challenges, we propose a throughput maximization algorithm to enhance network communication efficiency and design a winner and pricing determination algorithm to establish a reasonable market. The simulation results demonstrate that our mechanism achieves higher transaction quantity than benchmark schemes (McAfee, Candidate critical pair, and Vickrey-Clarke-Groves (VCG)), motivating UAVs to participate in relay tasks for users.

Drunk-driving is an important factor causing road traffic accidents and deaths, which deserves a lot of research. However, most current methods for detecting drunk-driving depend on customized hardware or require users’ active participation, making it impractical to monitor blood alcohol content (BAC) during driving. This article introduces BACFuse, a device-free, contactless, and noninvasive system utilizing smartphone in driving environments, which achieves relatively high accuracy in drunk-driving monitoring by integrating various voice sensing modalities. BACFuse first captures vocal cord vibration from ultrasonic signals, then records voice commands from audio signals. BACFuse combines the ultrasonic signals with audio signals and effectively detects drunk-driving and BAC. A key enabler lies in our modeling of latent interaction between acoustic and ultrasonic signals to mitigate ambient noise, realizing noise-resistant drunk-driving detection. Additionally, we propose an effective modules within the co-attention method to fuse the multimodal signals, further enhancing the accuracy of drunk-driving detection. We conduct extensive experiments to evaluate BACFuse’s performance on 20 participants in safe laboratory experiments. The results demonstrate that our system achieves BAC measurement with an MAE of 2.13 mg/dl, showing promise for future in-car driving management paradigms.

Recent studies have revealed the vulnerability of Deep Neural Network (DNN) models to backdoor attacks. However, existing backdoor attacks arbitrarily set the trigger mask or use a randomly selected trigger, which restricts the effectiveness and robustness of the generated backdoor triggers. In this paper, we propose a novel attention-based mask generation methodology that searches for the optimal trigger shape and location. We also introduce a Quality-of-Experience (QoE) term into the loss function and carefully adjust the transparency value of the trigger in order to make the backdoored samples to be more natural. To further improve the prediction accuracy of the victim model, we propose an alternating retraining algorithm in the backdoor injection process. The victim model is retrained with mixed poisoned datasets in even iterations and with only benign samples in odd iterations. Besides, we launch the backdoor attack under a co-optimized attack framework that alternately optimizes the backdoor trigger and backdoored model to further improve the attack performance. Apart from DNN models, we also extend our proposed attack method against vision transformers. We evaluate our proposed method with extensive experiments on VGG-Flower, CIFAR-10, GTSRB, CIFAR-100, and ImageNette datasets. It is shown that we can increase the attack success rate by as much as 82% over baselines when the poison ratio is low and achieve a high QoE of the backdoored samples. Our proposed backdoor attack framework also showcases robustness against state-of-the-art backdoor defenses.

Radar spoofing attacks mislead victim radars by injecting false information. Successful attacks require prior knowledge of the victim radar's mode and parameters, and existing works obtain this critical information with expensive equipment, e.g., software-defined radio or spectrum analyzer. In this paper, we proposeWaston, a low-cost system for radar mode detection and parameter estimation using commercial off-the-shelf (COTS) mmWave radars. To overcome the disadvantage of low sampling frequency of COTS mmWave radars, we design two special local signals to detect frequency points and spectral shapes for radar mode detection. We propose a novel parameter estimation algorithm to estimate frequency- and time-domain parameters for spoofing different radars. We have implemented a prototype on the TI AWR1843 platform and conducted extensive experiments to evaluate the performance ofWaston. Our experimental results demonstrate thatWastonachieves an accuracy of 100$\%$for mode detection and 99$\%$for parameter estimation. Furthermore, we demonstrate that the estimated parameters can be used to launch a successful spoofing attack against the victim radar.

Backdoor attacks can exploit vulnerabilities in the training process of Deep Neural Networks (DNNs), introducing hidden malicious functionality that can be activated by a specific input pattern. Existing defenses typically rely on the assumption of a significant difference between poisoned and clean samples. However, subtle differences in dynamic and low poisoning ratio attacks can conceal poisoning features, thereby evading defenses. In this work, we propose a novel backdoor defense approach calledArtemis, which utilizes distribution shifts to eliminate the discrepancy between poisoned and benign samples in the feature space. Additionally,Artemislearns from domain-invariant features after the shift. To further enhance the purification of backdoors in DNNs, we incorporate soft knowledge distillation intoArtemisto guide the alignment of features between the source dataset domain and the generated distribution-shift dataset domain. We extensively compare our proposed method with 5 state-of-the-art (SOTA) defensive techniques under 9 SOTA attacks on 4 datasets to demonstrate its effectiveness and robustness. Our results show that our method,Artemis, can successfully purify dynamic and low poisoning ratio backdoor attacks and outperform existing defenses by a significant margin. We release the code athttps://github.com/xmyun/Artemis.

As cloud storage becomes popular, more and more users tend to outsource their data to powerful cloud severs. To prevent a single point of failure, users prefer to store data on multiple cloud servers from different cloud service providers. However, after outsourcing data to cloud servers, users lose the control of their data, which may incur many serious security issues, such as abnormal data tampering and deleting. It is necessary for users to audit multiple cloud storage aperiodically. In this article, we aim to design a decentralized self-auditing solution for multiple cloud storage, in which cloud servers can audit the integrity of each other, and thus no third-party entity is required. First, based on basic algebra, our protocol realizes decentralized self-auditing multiple cloud storage that only involves encrypted user data. Second, we design a proof exchanging mechanism, making any number of cloud servers form the same final integrity proof with a linear number of interactions. Third, our solution can achieve cloud dynamics, which can freely enable and disable a cloud server to provide storage service. At last, security proof and performance evaluation show that the proposed protocol has provable security and high efficiency.

Watermark is essential for protecting the intellectual property of private images. However, a wide range of watermark removal attacks, especially many AI-powered ones, can automatically predict and remove watermarks, posing serious concerns. In this paper, we present the design ofImprints, a defensive watermarking framework that fortifies watermarks against watermark removal attacks. By formulating an optimization problem that deters watermark removal attacks, we design image-independent/dependent defensive watermark models for effective batch/customized protection. We further enhance the watermark to be transferable to unseen watermark removal attacks and robust to editing distortions. Extensive experiments verify thatImprintsoutperforms existing baselines in terms of its immunity to 8 state-of-the-art watermark removal attacks and 3 commercial black-box watermark removal software. The source code is available athttps://github.com/Imprints-wm/Imprints.

Existing research has demonstrated that deep neural networks are susceptible to model extraction attacks, where an attacker can construct a substitute model with similar functionality to the victim model by querying the black-box victim model. To counter such attacks, various disruption-based defenses have been proposed. These defenses disrupt the output results of queries before returning them to potential attackers. In this paper, we propose the first defense-penetrating model extraction attack framework, aimed at breaking disruption-based defense methods. Our proposed attack framework comprises two key modules: disruption detection and disruption recovery, which can be integrated into generic model extraction attacks. Specifically, the disruption detection module uses a novel meta-learning-based algorithm to infer the defense strategy employed by the defender, by learning the key differences between the distributions of disrupted and undisrupted query results. Once the defense method is inferred, the disruption recovery module is designed to restore clean query results from the disrupted query results, using a carefully-designed generative model. We conducted extensive experiments on 5 commonly-used datasets to evaluate the effectiveness of our proposed framework. The results demonstrate that the substitute model accuracy of current model extraction attacks can be significantly improved by up to 82.42%, even when faced with four state-of-the-art model extraction defenses. Moreover, our attack approach shows promising results in penetrating unknown defenses in real-world cloud service APIs hosted by Microsoft Azure and Face++.

Due to the widespread use of mobile devices, it is essential to authenticate users on mobile devices to prevent sensitive information leakage. Biometrics-based authentication is prevalent on smart devices to verify the legitimacy of users, but is vulnerable to replay attacks. In this paper, we propose to leverage the distinctive finger tap gesture during unlocking smartphone to establish a secure multi-factor authentication system, named FingerVib. Compared with other biometric-based authentication systems, FingerVib does not require users to remember any complicated information (e.g., hand gestures, doodles) and the working type is unobtrusive. When users unlock their phones by tapping, FingerVib utilizes the microphone to record the sound produced by fingers tapping on the phone and adopts IMU (Inertial Measurement Unit) to extract the vibration of users’ smartphones. One key contribution is that we model the inherent correlation between sounds and vibration signals. Specifically, FingerVib captures two novel reactions to describe how the individual’s contact palm modulates signals in two different domains. Based on these two responses, we develop a real-time noise-resistant unlocking activity detection algorithm, which allows accurate unlocking signal segmentation even if the two modalities are interfered. Further, we develop a modal fusion model where the model extracts cross-modal features and acquires inter-modal correlation features to ensure consistent performance of inference even when modalities are disturbed. In a user study with 41 participants, FingerVib achieves an authentication accuracy of 98.53% and an average performance of 1.36% FAR, 2.76% FRR and 2.72% EER against replay attacks and impersonation attacks. FingerVib’s fusion approach improves identification performance by roughly 9.7% and 11.6% over Wavocie and AUDIOIMU, respectively, within existing multi-modal fusion systems. Extensive experimental results demonstrate the effectiveness and robustness of FingerVib under various conditions.

Blood pressure (BP) is a critical physiological parameter closely associated with severe diseases such as heart failure and kidney damage. Current methods either require additional or dedicated hardware, or closing touching to the devices, causing discomfort and inconvenience. Therefore, a convenient, contactless BP measurement solution is highly desired. In this work, we present Ubicon-BP, a ubiquitous, device-free, and contactless BP detection application. Ubicon-BP calculates BP based on the pulse transit time (PTT), a key feature that is medically proven correlated with BP. However, using smartphone sensors to contactless calculate PTT is non-trivial since it requires a micro-second level precision for cardiac event detection. To address this issue, we propose leveraging the acoustic sensors in smartphone to detect vibrations caused by heart valve movements, as well as camera sensors to measure finger pulses. To accurately measure heartbeat signal that are susceptible to motion, we first improve the sensing granularity of acoustic signals and then introduce the IQ-MVED model to eliminate motion interference. Furthermore, when recovering pulse signals from video signals, issues such as poor generalization performance arise. Consequently, we propose the TS-CAN and meta-learning models to obtain personalized pulse signals. Finally, we transform the extracted time-frequency features from the recovered heartbeats and pulse signals to the corresponding BP. Comprehensive testing involving 50 subjects reveal a standard deviation error of$ 4.27 \; \text{mmHg}$for diastolic pressure and$ 6.36 \; \text{mmHg}$for systolic pressure, respectively.

Early detection of myocardial infarction (MI) is essential for alleviating symptoms and improving daily activity performance. Researchers typically employ continuous segments of heartbeat signals (20–30 seconds), such as ECG signals, for MI detection, as MI often induces changes in heartbeat patterns. Current MI detection methods, like wearable sensors, may induce discomfort from prolonged wear, and Radio Frequency (RF) based approaches might fail to extract fine-grained heartbeat signals during vigorous movement. This article presents a reliable and motion-robust MI detection method based on RF signals. By developing a series of advanced signal processing algorithms, MI-Ra can capture fine-grained heartbeat signals during various daily activities. Our design is inspired by the fact that RF reflections caused by heartbeat signals are mixed with other motion-induced reflections in a nonlinear manner. We utilize the Taylor series expansion method to extract the linear component of these mixed non-linear signals and propose a novel Generative Adversarial Networks (GAN) method, named IQ-TransGAN, to separate the heartbeat signal. To enhance MI detection reliability, MI-Ra employs a multi-periodicity modeling method to extract refined signal representations from recovered heartbeat signals. We have recruited 50 volunteers with MI from Zhongnan Hospital of Wuhan, China, and 50 volunteers without MI, for comprehensive evaluations. The results demonstrate that MI-Ra achieves an average MI detection accuracy of 95.2% when user is quasi-stationary. Even during user non-stationary conditions, MI-Ra maintains an average detection accuracy of 90.5%. MI-Ra shows promise in paving way for smart home healthcare.

Machine learning models have become integral to various aspects of daily life, prompting increased vulnerability to adversarial attacks.Adversarial training is one of the most promising and practical methods to enhance model robustness.Existing adversarial training methods, however, assume access to the original training data.But nowadays, more and more users directly download models from the open-source model platforms or tech companies, but the original training datasets are usually unreleased because of commercial interests or privacy.In such scenarios, the user cannot utilize the former adversarial training methods to improve model robustness because of the lack of original training datasets.Thus, we present the first exploration of a data-free adversarial training framework, Alchemy, which seeks to enhance model robustness without requiring access to the original training data.By addressing the notable challenges of reconstructing high-quality training data with robust features and improving the adversarial robustness to the inaccessible original dataset, our approach achieves the goals of both high accuracy maintenance and robustness improvement.Comprehensive experiments on four datasets compared with five baselines, demonstrate Alchemy 's high effectiveness.With no access to any training dataset, the average robustness improvement with Alchemy is effective in most attack scenarios.Additional evaluations underscore the framework's stability under different settings and discuss future research directions.

Machine Learning as a Service (MLaaS) enables resource-constrained users to access well-trained models through a publicly accessible Application Programming Interface (API) on a pay-per-query basis.Nevertheless, model owners may face the potential threats of model extraction attacks where malicious users replicate valuable commercial models based on query results.Existing defenses against model extraction attacks, however, either sacrifice prediction accuracy or fail to thwart more advanced attacks.In this paper, we propose a novel model extraction defense, dubbed Beowulf 1 , which draws inspiration from theoretical findings that models with complex and narrow decision regions are difficult to be reproduced.Rather than arbitrarily altering decision regions, which may jeopardize the predictive capacity of the victim model, we introduce a dummy class, carefully synthesized using both random and adversarial noises.The random noise broadens the coverage of the dummy class, and the adversarial noise impacts decision regions near decision boundaries with normal classes.To further improve the model utility, we propose to employ data augmentation methods to seamlessly integrate the dummy class and the normal classes.Extensive evaluations on CIFAR-10, GTSRB, CIFAR-100, and ImageNette datasets * Yanjiao Chen and Qian Wang are corresponding authors.

Text-to-image (T2I) models, such as Stable Diffusion, have exhibited remarkable performance in generating high-quality images from text descriptions in recent years. However, text-to-image models may be tricked into generating not-safe-for-work (NSFW) content, particularly in sexually explicit scenarios. Existing countermeasures mostly focus on filtering inappropriate inputs and outputs, or suppressing improper text embeddings, which can block sexually explicit content (e.g., naked) but may still be vulnerable to adversarial prompts -- inputs that appear innocent but are ill-intended. In this paper, we present SafeGen, a framework to mitigate sexual content generation by text-to-image models in a text-agnostic manner. The key idea is to eliminate explicit visual representations from the model regardless of the text input. In this way, the text-to-image model is resistant to adversarial prompts since such unsafe visual representations are obstructed from within. Extensive experiments conducted on four datasets and large-scale user studies demonstrate SafeGen's effectiveness in mitigating sexually explicit content generation while preserving the high-fidelity of benign images. SafeGen outperforms eight state-of-the-art baseline methods and achieves 99.4% sexual content removal performance. Furthermore, our constructed benchmark of adversarial prompts provides a basis for future development and evaluation of anti-NSFW-generation methods.

Given the societal impact of unsafe content generated by large language models (LLMs), ensuring that LLM services comply with safety standards is a crucial concern for LLM service providers. Common content moderation methods are limited by an effectiveness-and-efficiency dilemma, where simple models are fragile while sophisticated models consume excessive computational resources. In this paper, we reveal for the first time that effective and efficient content moderation can be achieved by extracting conceptual features from chat-oriented LLMs, despite their initial fine-tuning for conversation rather than content moderation. We propose a practical and unified content moderation framework for LLM services, named Legilimens, which features both effectiveness and efficiency. Our red-team model-based data augmentation enhances the robustness of Legilimens against state-of-the-art jailbreaking. Additionally, we develop a framework to theoretically analyze the cost-effectiveness of Legilimens compared to other methods

Membership inference attacks have been proposed to infer whether a specific sample is in the training dataset of a victim model. Inferred membership may reveal sensitive information, e.g., personal health condition deduced from a disease prediction model. In federated learning where local datasets of different participants are supposed to be protected, membership inference attacks may still pose a privacy threat. However, existing membership inference attacks in federated learning select the final epoch and random intermediate epochs during training to solicit features for the attack, which may lead to poor attack performance. In this paper, we propose a novel membership inference attack in federated learning, which attempt to find the key epoch during training that is most indicative of the differences between member and non-member samples. Inspired by the overfitting phenomena that often occurs during the training of learning models, we derive the key epoch as the onset of overfitting. We design efficient algorithms to pinpoint the key epoch and leverage attention mechanism to weight the importance of different features. We evaluate our scheme on four datasets and compare experimental results with state-of-the-art attack algorithms. Experimental results demonstrate that our attack outperforms existing works in terms of inference accuracy.

Instead of building deep learning models from scratch, developers are more and more relying on adapting pre-trained models to their customized tasks. However, powerful pre-trained models may be misused for unethical or illegal tasks, e.g., privacy inference and unsafe content generation. In this paper, we introduce a pioneering learning paradigm, non-fine-tunable learning, which prevents the pre-trained model from being fine-tuned to indecent tasks while preserving its performance on the original task. To fulfill this goal, we propose Sophon, a protection framework that reinforces a given pre-trained model to be resistant to being fine-tuned in pre-defined restricted domains. Nonetheless, this is challenging due to a diversity of complicated fine-tuning strategies that may be adopted by adversaries. Inspired by model-agnostic meta-learning, we overcome this difficulty by designing sophisticated fine-tuning simulation and fine-tuning evaluation algorithms. In addition, we carefully design the optimization process to entrap the pre-trained model within a hard-to-escape local optimum regarding restricted domains. We have conducted extensive experiments on two deep learning modes (classification and generation), seven restricted domains, and six model architectures to verify the effectiveness of Sophon. Experiment results verify that fine-tuning Sophon-protected models incurs an overhead comparable to or even greater than training from scratch. Furthermore, we confirm the robustness of Sophon to three fine-tuning methods, five optimizers, various learning rates and batch sizes. Sophon may help boost further investigations into safe and responsible AI.

The interaction mode of smart eyewear has garnered significant research attention. Most smart eyewear relies on touchpads for user interaction. This article identifies a drawback arising from the use of touchpads, which can be obtrusive and unfriendly to users. In this article, we propose TeethFa, a novel fabric sensor-based system for recognizing teeth gestures. TeethFa serves as a hands-free interaction method for smart eyewear. TeethFa utilizes fabric sensors embedded in the glasses frame to capture pressure changes induced by facial muscle movements linked to teeth movements. This enables the identification of subtle teeth gestures. To detect teeth gestures, TeethFa designs a novel template-based signal segmentation method to determine the boundary of teeth gestures from fabric sensors, even in the presence of motion interference. To improve TeethFa’s generalization, we employ a meta-learning technique based on generalization adjustment to extend the model to new users. We conduct extensive experiments to assess TeethFa’s performance on 30 volunteers. The results demonstrate that our system accurately identifies five different teeth gestures with an average accuracy of 93.57%, and even for new users, the accuracy can reach 89.58%. TeethFa shows promise in offering a new interaction paradigm for smart eyewear in the future.

The breathtaking progress in machine learning has motivated studies in learning-based congestion control algorithms, which are expected to adjust congestion window choices according to the dynamic network environment. Nevertheless, existing learning-based congestion control protocols are mostly designed and trained for a specific network environment. When applied to a different network environment, the previously trained model may see a considerable degradation in performance. As the rapid development of communication technologies has given rise to the emergence of a diversity of new networks, it is desirable for learning-based congestion control models to quickly transfer to different network environments. Driven by this motivation, we propose a novel TRansferable Congestion Control (TRCC) protocol, which takes full advantage of both reinforcement learning and transfer learning to intelligently cope with network congestion in scenarios. The key idea to enable a fast transfer from the source network environment to the target network environment is to fine-tune a well-trained model in the source network to suit the target network in a time-effective way. We theoretically prove the transferability and quick convergence of our proposed transfer reinforcement learning-based congestion control algorithm by deriving the Markov transition matrix and the similarity of the reward function. Our experiments validate that TRCC can converge in a new network environment in a short time while achieving comparable performance with baseline algorithms.

Deep neural networks are vulnerable to backdoor attacks, where a specially-designed trigger will lead to misclassification of any benign samples. However, existing backdoor attacks usually impose conspicuous patch triggers on images, which are easily detected by humans and defense algorithms. Existing works on invisible triggers, however, either have reduced attack success rate or yield detectable patterns to visual inspections. In this paper, we proposeKerbNet, a kernel-based backdoor attack framework, which applies kernel operations to clean samples as the trigger to incur misclassification. The kernel-processed samples achieve a high attack success rate while appearing natural with high Quality-of-Experience (QoE). We carefully design the kernel trigger generation algorithm by exploiting the neural network structure to propagate the influence of the trigger to the target misclassification label under the QoE constraint. We conduct extensive experiments on five datasets, i.e., MNIST, GTSRB, CIFAR-10, CelebA, and ImageNette to evaluate the effectiveness and practicality ofKerbNetunder the impact of various factors, including neuron-residing layer, kernel size, base image, loss function, model structure, and so on. We also show that our proposed attacks can evade state-of-the-art defense strategies and visual inspections. Code will be available after publication.

Backdoor attacks have been widely studied for image classification tasks, but rarely investigated for video recognition tasks. In this paper, we explore the possibility of physically-realizable backdoor attacks against video recognition models. Different from existing works that directly apply image backdoor attacks to videos, i.e., patch a visible trigger to each frame of a video, we carefully take into consideration the temporal interactions among frames in a video. Our proposed video backdoor attack, namedPalette, features two special design choices. The first is to utilize natural-light-alike RGB offset as triggers rather than traditional patch triggers. Such triggers may be applied in the physical world through lighting without the need to modify video files. The second is to make the backdoored model more robust to temporal asynchronization between the trigger and the video samples by performing rolling operations during sample poisoning. Extensive experiments show thatPaletteoutperforms existing video backdoor attacks, especially in the physical world. It is shown thatPaletteis also resistant to backdoor defense methods. We will open-source our codes upon publication.

When it comes to the marriage of federated learning and personalized recommendation services (FedRec), characterizing user-item interaction behaviors is a long-standing and unresolved issue, highlighting the growing data privacy concerns due to the inherent openness of recommender systems. As the new interaction-level membership inference attacks on FedRecs have recently surfaced, quite possibly such adversarial attacks act as the hidden confounders lying behind the interactive recommendation, resulting in the obstruction of the causal effect disentanglement on long-term user satisfaction. As such, tailored to the specifics of private learning, we propose a counterfactual interactive recommendation system that builds a differentially private representation learning based defender (CIRDP) to capture and mitigate the adversarial threats, augmenting causal inference-based interactive recommendation of FedRecs. When characterizing interaction-level membership inference attacks of the hidden eavesdropping adversary as the primary cause of adversarial effect on user satisfaction, CIRDP incorporates causal inference-augmented offline reinforcement learning (offline RL) into FedRecs. CIRDP innovatively provides counterfactual satisfaction by optimizing a sensitivity-guided disentangled representation module with an innovative two-fold mutual information objective. As such, CIRDP introduces a differentially private representation learning based defender, guaranteeing interaction behavior-level differential privacy (DP) with a significant reduction in privacy costs. Extensive comparisons demonstrate CIRDP’s superiority over the state-of-the-art baselines in reducing inference attack threats and improving long-term success in the interactive recommendation.

Traffic anomaly detection in road networks is vital for patrol participants to adaptively optimize patrol routes for hybrid task scheduling. In most cases, the routine patrol routing problem comes down to offline task scheduling with a maximal coverage requirement of traffic violation hotspots. Meanwhile, emergent anomalies (e.g., accidents, crowded events) can happen with the unpredictable duration times and locations, leading to real-time requirements of online task scheduling. In practice, trading off coverage and emergency is usually inapplicable in hybrid scenarios, especially with decentralized routing decision makers in highly stochastic traffic environments. In this paper, we propose a Crowd-enhanced hybrid task scheduling framework with the balance of Coverage and Emergency, calledCCE-patrol. CCE-patrol consists of offline violation hotspot patrolling, online emergency monitoring. For the offline subproblem, a contact-based integer linear problem modeling is proposed to adaptively find optimal patrol routes. Guided by offline routing, we identify and incorporate online emergent anomalies into traffic data completion for urban crowdsensing, with up-to-date importance estimation by multi-agent cooperative reinforcement learning. The secretary problem can be utilized to select important monitoring tasks in an online manner. Extensive experiments show the superiority of our proposals in comparison with baseline algorithms (i.e., HASH (Sun et al. 2022)) by 21.08$\%$improvement.

Tracking the menstrual cycle (MC) is essential for women to manage their health and schedule, especially for those with irregular MC. Existing MC tracking methods either rely on length of previous cycles (e.g., calendar noting) or require additional devices to collect more information (e.g., basal temperature), which are not able to realize both accuracy and convenience. Inspired by the medical studies that gait patterns will be affected by MC, we design a smartphone-based application named MC-Tracking, which monitors MC based on the Inertial Measurement Unit (IMU) signals. By identifying the walking activity based on the acceleration and angular velocity signals, we train an attention-based prediction model that can be generalized to new users with meta learning. 40 volunteers participate in an extensive experiment for more than 3 months, in which more than 2.4 TB of time-series data is collected to evaluate the performance of MC-tracking. It is verified that MC-tracing can predict the onset of MC seven days in advance with an average error of 0.56 days. We also demonstrate that the prediction accuracy is robust to the age, emotion, biological clock and smartphone brand.

Long-term incorrect sitting undoubtedly will damage physical health. Recognizing bad sitting posture has been of particular interest recently due to the prevailing Internet of Healthcare Things (IoHT). While various sitting posture recognition systems based on wearable devices and cameras are designed, they expose two obvious weaknesses. First, the sensors attached to the body will cause inconvenience to users, and using a camera requires high energy consumption and faces the risk of user privacy leakage. Second, most of these systems require massive training samples to build models, and the recognition performance of certain models on new user data with significant sample distribution differences remains poor. In this work, we propose SmartSit, the first-ever robust sitting posture recognition system with smartphone acoustic sensing. We start by designing a signal detection algorithm to determine the boundary of the sitting posture signal through a series of signal transformation methods. Then we construct the sitting posture recognition module MG-Reptile by modifying the meta-learning method by combining the Distributed Measurement Strategy (DMS) and Generative Adversarial Network (GAN). We show that the designed system is immune to the low generalization performance with only a few training samples. The observed testing results further validate the effectiveness and robustness of SmartSit.

The rising demand for utilizing fine-grained data in deep-learning (DL) based intelligent systems presents challenges for the collection and transmission abilities of real-world devices. Deep compressive sensing, which employs deep learning algorithms to compress signals at the sensing stage and reconstruct them with high quality at the receiving stage, provides a state-of-the-art solution for the problem of large-scale fine-grained data. However, recent works have proven that fatal security flaws exist in current deep learning methods and such instability is universal for DL-based image reconstruction methods. In this article, we assess the security risks introduced by deep compressive sensing in the widely used computer vision system in the face of adversarial example attacks and poisoning attacks. To implement the security inspection in an unbiased and complete manner, we develop a comprehensive methodology and a set of evaluation metrics to manage all potential combinations of attack methods, datasets (application scenarios), categories of deep compressive sensing models, and image classifiers. The results demonstrate that deep compressive sensing models unknown to adversaries can protect the computer vision system from adversarial example attacks and poisoning attacks, whereas the ones exposed to adversaries can cause the system to become more vulnerable.

Alzheimer’s disease (AD) is an insidious and progressive neurodegenerative disease, and the annual relevant social cost for AD patients can reach about $1 trillion worldwide. Therefore, early diagnosis and treatment of AD play a vital role in slowing disease progression. However, existing detection methods for cognitive impairment cannot consistently screen the stage of AD. To tackle this challenge, we propose an AD detection system, Ubi-AD, which combines the features of multiple biomarkers to realize passive and accurate AD detection. Unlike existing work, Ubi-AD can passively recognize the AD digital biomarkers during daily smartwatch usage without interfering with the user. At the user end, Ubi-AD first extracts the non-speech sounds (pause words, such as em, ah), which contain no privacy-sensitive content. Then, Ubi-AD recognizes the user’s walking activity, dining activity, and sleep activity from daily activities. Ubi-AD analyzes these data from smartwatch and predicts the AD stages using a multi-modal fusion neural network at the cloud end. We evaluate our model on a collected dataset from 45 volunteers. As a result, Ubi-AD can reach a detection accuracy of 93.4%, which means that Ubi-AD can provide multiple effective biomarkers for ubiquitous and passive detection in daily life.

Graph neural networks (GNNs) have been developed to mine useful information from graph data of various applications, e.g., healthcare, fraud detection, and social recommendation. However, GNNs open up new attack surfaces for privacy attacks on graph data. In this paper, we propose Infiltrator, a privacy attack that is able to pry node-level private information based on black-box access to GNNs. Different from existing works that require prior information of the victim node, we explore the possibility of conducting the attack without any information of the victim node. Our idea is to infiltrate the graph with attacker-created nodes to befriend the victim node. More specifically, we design infiltration schemes that enable the adversary to infer the label, neighboring links, and sensitive attributes of a victim node. We evaluate Infiltrator with extensive experiments on three representative GNN models and six real-world datasets. The results demonstrate that Infiltrator can achieve an attack performance of more than 98% in all three attacks, outperforming baseline approaches. We further evaluate the defense resistance of Infiltrator against the graph homophily defender and the differentially private model.

Federated learning is exposed to model poisoning attacks as compromised clients may submit malicious model updates to pollute the global model. To defend against such attacks, robust aggregation rules are designed for the centralized server to winnow out outlier updates, and to significantly reduce the effectiveness of existing poisoning attacks. In this paper, we develop an advanced model poisoning attack against defensive aggregation rules. In particular, we exploit the catastrophic forgetting phenomenon during the process of continual learning to destroy the memory of the global model. Our proposed framework, called Oblivion, features two special components. The first component prioritizes the weights that have the most influence on the model accuracy for poisoning, which induces a more significant degradation on the global model than equally perturbing all weights. The second component smooths malicious model updates based on the number of selected compromised clients in the current round, adjusting the degree of poisoning to suit the dynamics of each training round. We implement a fully-functional prototype of Oblivion in PLATO, a real-world scalable federated learning framework. Our extensive experiments over three datasets demonstrate that Oblivion can boost the attack performance of model poisoning attacks against unknown defensive aggregation rules.

Recent studies show that machine learning models are vulnerable to model extraction attacks, where the adversary builds a substitute model that achieves almost the same performance of a black-box victim model simply via querying the victim model. To defend against such attacks, a series of methods have been proposed to disrupt the query results before returning them to potential attackers, greatly degrading the performance of existing model extraction attacks.In this paper, we make the first attempt to develop a defense-penetrating model extraction attack framework, named D-DAE, which aims to break disruption-based defenses. The linchpins of D-DAE are the design of two modules, i.e., disruption detection and disruption recovery, which can be integrated with generic model extraction attacks. More specifically, after obtaining query results from the victim model, the disruption detection module infers the defense mechanism adopted by the defender. We design a meta-learning-based disruption detection algorithm for learning the fundamental differences between the distributions of disrupted and undisrupted query results. The algorithm features a good generalization property even if we have no access to the original training dataset of the victim model. Given the detected defense mechanism, the disruption recovery module tries to restore a clean query result from the disrupted query result with well-designed generative models. Our extensive evaluations on MNIST, FashionMNIST, CIFAR-10, GTSRB, and ImageNette datasets demonstrate that D-DAE can enhance the substitute model accuracy of the existing model extraction attacks by as much as 82.24% in the face of 4 state-of-the-art defenses and combinations of multiple defenses. We also verify the effectiveness of D-DAE in penetrating unknown defenses in real-world APIs hosted by Microsoft Azure and Face++.

Recent works have revealed the vulnerability of deep neural networks to backdoor attacks, where a backdoored model orchestrates targeted or untargeted misclassification when activated by a trigger. A line of purification methods (e.g., fine-pruning, neural attention transfer, MCR [69]) have been proposed to remove the backdoor in a model. However, they either fail to reduce the attack success rate of more advanced backdoor attacks or largely degrade the prediction capacity of the model for clean samples. In this paper, we put forward a new purification defense framework, dubbed SAGE, which utilizes self-attention distillation to purge models of backdoors. Unlike traditional attention transfer mechanisms that require a teacher model to supervise the distillation process, SAGE can realize self-purification with a small number of clean samples. To enhance the defense performance, we further propose a dynamic learning rate adjustment strategy that carefully tracks the prediction accuracy of clean samples to guide the learning rate adjustment. We compare the defense performance of SAGE with 6 state-of-the-art defense approaches against 8 backdoor attacks on 4 datasets. It is shown that SAGE can reduce the attack success rate by as much as 90% with less than 3% decrease in prediction accuracy for clean samples. We will open-source our codes upon publication.

Federated learning (FL) has received great attention in recent years due to its good performance and privacy security. However, there are still some problems in FL, such as resource allocation, client selection and incentive mechanism. Auctions, as an incentive mechanism with many advantages, solve these problems well. In this paper, we propose DDS, a lightweight and effective method to evaluate the data value for the scenario of federated learning. Based on DDS, we can pick up clients with high training value with low calculation cost. Furthermore, our method is independent of data aggregation, which makes it possible to implement in federated learning. We also conduct the simulation experiments to demonstrate the effectiveness and robustness of our method.

Recently more and more cloud service providers (e.g., Microsoft, Google, and Amazon) have commercialized their well-trained deep learning models by providing limited access via web API interfaces. However, it is shown that these APIs are susceptible to model inversion attacks, where attackers can recover the training data with high fidelity, which may cause serious privacy leakage.Existing defenses against model inversion attacks, however, hinder the model performance and are ineffective for more advanced attacks, e.g., Mirror [4]. In this paper, we proposed NetGuard, a novel utility-aware defense methodology against model inversion attacks (MIAs). Unlike previous works that perturb prediction outputs of the victim model, we propose to mislead the MIA effort by inserting engineered fake samples during the training process. A generative adversarial network (GAN) is carefully built to construct fake training samples to mislead the attack model without degrading the performance of the victim model. Besides, we adopt continual learning to further improve the utility of the victim model. Extensive experiments on CelebA, VGG-Face, and VGG-Face2 datasets show that NetGuard is superior to existing defenses, including DP [37] and Ad-mi [32] on state-of-the-art model inversion attacks, i.e., DMI [8], Mirror [4], Privacy [12], and Alignment [34].

The energy harvesting of unmanned aerial vehicle (UAV) has been researched extensively in recent years. However, the existing energy harvesting between the base station and UAVs does not consider the information asymmetry factor, which means the base station provides the radio frequency (RF) energy for UAVs in the context of UAVs’ partial private information. In order to maximize the base station’s utility or payoff, it is crucial for the base station to motivate more UAVs to harvest RF energy. In the paper, we propose an effective incentive energy harvesting mechanism in UAV communication network, which is a challenging problem since there exist interest conflicts that the base station and UAVs are rational individuals who maximize their utilities. Our objective is to make the base station’s utility maximum via balancing the tradeoff between transmit power cost and charged price benefit, while incentivizing UAVs to purchase transmit power. We design a series of optimal energy harvesting contract with different price discounts targeting different types of UAVs by contract theory. Owing to information asymmetry, we analyze two different information scenarios: complete and incomplete information. We suppose the base station knows each UAV’s type in complete information, then we analyze the practical case that the base station is aware of incomplete information of UAV’s private information. The base station aims to maximize its utility by providing contract. The UAVs choose the contract meeting the individual rationality (IR) and incentive compatibility (IC) rules while maximizing their utilities. Our simulation shows that the energy harvesting mechanism maximizes the base station’s utility and stimulates UAVs to purchase RF energy transmit power in different scenarios. Compared with other methods, our proposed optimal contract can improve the utility of the base station while maximizing the utility of UAVs.

The rapid development of the Internet of Things (IoT) boosts the spread of intelligent spaces. Biometrics-based user identification has gained great popularity recently, among which gait analysis offers a stable, user-friendly, and economical solution. Thanks to the advancement in wireless sensing technologies, capturing gait characteristics using WiFi signals has become a promising new paradigm. The identification process is contactless, insensitive to lighting conditions, and can reuse the incumbent WiFi infrastructure. In this article, we present a gait-based dual-user identification framework named WiWalk to tackle the difficulty where users walk closely together with mixed effects on WiFi signals. The core of WiWalk is to train a deep neural network that can separate and recover individual signals from the mixed ones. Since the separation process inevitably causes information loss, we carefully design a series of algorithms for interference elimination, segmentation, and feature extraction, to enhance the identification accuracy. We conduct extensive experiments to evaluate WiWalk at different locations and times with users of different ages, genders, clothing, and walking behaviors. WiWalk can reach an accuracy of 94.44%, which is suitable for smart homes or offices with a small user base.

Asthma is a common respiratory disease in modern society. However, people are rarely aware of the symptoms of asthma because the early stage of asthma is similar to that of the common cold (e.g., wheeze, cough, and shortness of breath). To tackle this challenge, we propose an asthma detection system, Ubi-Asthma, based on the smartwatch. Ubi-Asthma combines breathing signals and guttural sound (e.g., cough sound and throat-clearing sound) signals to realize passive and accurate asthma detection without interrupting the user. Not only can Ubi-Asthma extract breathing signals from the user even if the user is walking but also recognize guttural sound signals when the user is engaged in voice communication without being disturbed by noise. The features of breathing and guttural sound are combined to improve the accuracy of asthma detection. We have implemented a fully functional prototype using an off-the-shelf smartwatch. Fifty volunteers participate in an extensive experiment of up to 150 h to train Ubi-Asthma. As a result, Ubi-Asthma can reach a detection accuracy of 98.4%, which is higher than breath- or guttural-based asthma detection systems. Ubi-Asthma is expected to provide a potential solution for smart-home applications in the future.

Recent works have revealed that backdoor attacks against Deep Reinforcement Learning (DRL) could lead to abnormal action selections of the agent, which may result in failure or even catastrophe in crucial decision processes. However, existing attacks only consider single-agent reinforcement learning (RL) systems, in which the only agent can observe the global state and have full control of the decision process. In this article, we explore a new backdoor attack paradigm in cooperative multi-agent reinforcement learning (CMARL) scenarios, where a group of agents coordinate with each other to achieve a common goal, while each agent can only observe the local state. In the proposed MARNet attack framework, we carefully design a pipeline of trigger design, action poisoning, and reward hacking modules to accommodate the cooperative multi-agent settings. In particular, as only a subset of agents can observe the triggers in their local observations, we maneuver their actions to the worst actions suggested by an expert policy model. Since the global reward in CMARL is aggregated by individual rewards from all agents, we propose to modify the reward in a way that boosts the bad actions of poisoned agents (agents who observe the triggers) but mitigates the influence on non-poisoned agents. We conduct extensive experiments on three classical CMARL algorithms VDN, COMA, and QMIX, in two popular CMARL games Predator Prey and SMAC. The results show that the baselines extended from single-agent DRL backdoor attacks seldom work in CMARL problems while MARNet performs well by reducing the utility under attack by nearly 100%. We apply fine-tuning as a potential defense against MARNet and demonstrate that fine-tuning cannot entirely eliminate the effect of the attack.

Recent research has shown that deep neural networks are vulnerable to backdoor attacks. A carefully-designed backdoor trigger will mislead the victim model to misclassify any sample with the trigger to the target label. Nevertheless, existing works usually utilize visible triggers, such as a white square at the corner of the image, which are easily detected by human inspections. Current efforts on developing invisible triggers yield low attack success in the physical domain. In this paper, we propose Kaleidoscope, an RGB (red, green, and blue) filter-based backdoor attack method, which utilizes RGB filter operations as the backdoor trigger. To enhance the attack success rate, we design a novel model-dependent filter trigger generation algorithm. We also introduce two constraints in the loss function to make the backdoored samples more natural and less distorted. Extensive experiments on CIFAR-10, CIFAR-100, ImageNette, and VGG-Flower have demonstrated that RGB filter-processed samples not only achieve high attack success rate but also are unnoticeable to humans. It is shown that Kaleidoscope can reach an attack success rate of more than 84% in the physical world under different lighting intensities and shooting angles. Kaleidoscope is also shown to be robust to state-of-the-art backdoor defenses, such as spectral signature, STRIP, and MNTD.

With the development of deep learning, deep neural network (DNN)-based application have become an indispensable aspect of daily life. However, recent studies have shown that these well-trained DNN models are vulnerable to model inversion attacks (MIAs), where attackers can recover their training data with high fidelity. Although several defensive strategies have been proposed to mitigate the impact of such attacks, existing defenses will inevitably compromise the model performance and are ineffective against more sophisticated attacks, such as Mirror [5]. In this paper, we introduce a novel GAN-based defense approach against model inversion attacks. Unlike previous works that perturb the prediction vector of the model, we manipulate the training procedure of the victim model by incorporating carefully-designed GAN-based fake samples. We also adjust the loss of the inversed samples to inject misleading features into the protected label of the victim model. Additionally, we adopt the concept of continual learning to improve the utility of the model. Extensive experiments conducted on the CelebA, VGG-Face, and VGG-Face2 datasets demonstrate that our proposed method outperforms existing defenses against state-of-the-art model inversion attacks, including DMI [9], Mirror [5], Privacy [13], and AMI [42]. It is shown that our proposed method can also retain a high defense performance in black-box scenarios.

Millimeter-wave (mmWave) radars have found applications in a wide range of domains, including human tracking, health monitoring, and autonomous driving, for their unobtrusive nature and high range accuracy. These capabilities, however, if used for malicious purposes, could also result in serious security and privacy issues. For example, a user’s daily life could be secretly monitored by a spy radar. Hence, there is a strong urge to develop systems that can detect and locate such spy radars. In this paper, we proposeRadar2, a practical system for passive spy radar detection and localization using a single commercial off-the-shelf (COTS) mmWave radar. Specifically, we propose a novelFrequency Component Detectionmethod to detect the existence of mmWave signals, distinguish between mmWave radar and WiGig signals using a waveform classifier based on a convolutional neural network (CNN), and localize spy radars using triangulation based on the detector’s observations at multiple anchor points. Not only doesRadar2work for different types of mmWave radar, but it can also detect and localize multiple radars simultaneously. Finally, we performed extensive experiments to evaluate the effectiveness and robustness ofRadar2in various settings. Our evaluation results show that the radar detection rate is above 96% and the localization error is within 0.3m. The results also reveal thatRadar2is robust against various environmental factors (e.g., room layout and human activities).

Social status, the social influence of a user, plays an important role in many real-world applications, e.g., trust relations and information propagation in a social network. In this paper, we reveal the possibility of falsifying social status through adversarial attacks in graph neural networks (GNNs). Different from neural networks in the visual or speech domain, GNNs take the attributes of nodes and edges in a graph as features. To cater to the characteristics of GNNs,$\vphantom {_{\int }}$we design a new paradigm of adversarial example attack, named$A^{2} S^{2}$- GNN ($\mathbf {GNN}$-based$\mathbf {A}$dversarial$\mathbf {A}$ttacks on$\mathbf {S}$ocial$\mathbf {S}$tatus), aiming at manipulating the social status of a target node in social networks. The key idea is to establish relationships or break relationships between a set of compromised nodes and the target node. More specifically, we consider a signed directed graph representing complicated positive/negative asymmetric relationships between nodes. We design an efficient adversarial attack algorithm to determine the minimum set of signed links that should be created or deleted to reach the attack objective. We conduct extensive experiments on baseline datasets. Compared with the benchmark algorithms,$A^{2} S^{2}$- GNN can effectively promote or vilify the social status of the target node up to 89.36% and 192.38%, respectively, while keeping the modification to the social network to the minimum. Furthermore, the experimental results on six status evaluating algorithms verify the transferability of our proposed attack algorithm.

With the unprecedented development of deep learning, autonomous vehicles (AVs) have achieved tremendous progress nowadays. However, AV supported by DNN models is vulnerable to data poisoning attacks, hindering the large-scale application of autonomous driving. For example, by injecting carefully designed poisons into the training dataset of the DNN model in the traffic sign recognition system, the attacker can mislead the system to make targeted misclassification or cause a reduction in model classification accuracy. In this article, we conduct a thorough investigation of the state-of-the-art data poisoning attacks and defenses against AVs. According to whether the attacker needs to manipulate the data labeling process, we divide the state-of-the-art attack approaches into two categories, i.e., dirty-label attacks and clean-label attacks. We also differentiate the existing defense methods into two categories based on whether to modify the training data or the models, i.e., data-based defenses and model-based defenses. In addition to a detailed review of attacks and defenses in each category, we also give a qualitative comparison of the existing attacks and defenses. Besides, we provide a quantitative comparison of the existing attack and defense methods through experiments. Last but not least, we pinpoint several future directions for data poisoning attacks and defenses in AVs, providing possible ways for further research.

Backdoor attacks aim to inject backdoors to victim machine learning models during training time, such that the backdoored model maintains the prediction power of the original model towards clean inputs and misbehaves towards backdoored inputs with the trigger. The reason for backdoor attacks is that resource-limited users usually download sophisticated models from model zoos or query the models from MLaaS rather than training a model from scratch, thus a malicious third party has a chance to provide a backdoored model. In general, the more precious the model provided (i.e., models trained on rare datasets), the more popular it is with users. In this article, from a malicious model provider perspective, we propose a black-box backdoor attack, named B 3 , where neither the rare victim model (including the model architecture, parameters, and hyperparameters) nor the training data is available to the adversary. To facilitate backdoor attacks in the black-box scenario, we design a cost-effective model extraction method that leverages a carefully constructed query dataset to steal the functionality of the victim model with a limited budget. As the trigger is key to successful backdoor attacks, we develop a novel trigger generation algorithm that intensifies the bond between the trigger and the targeted misclassification label through the neuron with the highest impact on the targeted label. Extensive experiments have been conducted on various simulated deep learning models and the commercial API of Alibaba Cloud Compute Service. We demonstrate that B 3 has a high attack success rate and maintains high prediction accuracy for benign inputs. It is also shown that B 3 is robust against state-of-the-art defense strategies against backdoor attacks, such as model pruning and NC.

Parkinson’s disease (PD) is a neurodegenerative disorder that severely affects the motor system of patients. Early PD detection will greatly improve the quality of lives. However, existing automatic PD detection systems either rely on customized sensors or require users to perform special activities, using machine learning models whose prediction process is not understandable by medical professionals. In this article, we develop a non-disruptive PD detection system on smartphones based on interpretable prediction models. We design an application named PDGes to passively collect touchscreen and Inertial Measurement Unit data of users’ tapping and swiping actions on smartphones. Meaningful features that reflect finger dexterity , tremor , stiffness , and hand movement are extracted to build the prediction model. To better comprehend the decisions made by the model, we conduct a systematic analysis of feature importance to help validate the conformity of the model with clinical PD diagnosis. We collected data from 108 volunteers to evaluate the performance of PDGes . The experiment results show that PDGes achieves a detection accuracy of more than 94.5% on different smartphones.

Age-of-Information (AoI) has been recently reckoned as a suitable parameter to evaluate the freshness of collected information, which is essential for data retrieval in Internet of Things, especially the monitoring tasks, e.g., the operating situation of equipments. To motivate a large number of sensor nodes and solicit more up-to-date information from these nodes, the control center usually allocates rewards to nodes according to their proportional contributions. This induces intense competitions among nodes who try to gain high payoffs by carefully balancing the rewards and the costs. In this article, we propose a novel stochastic game model to formulate the competition among sensor nodes, which considers AoI as a metric used by the control center to quantify the contributions of nodes. We also take into account the uncertainty of channel quality, which affects the transmission success ratio of packets generated by nodes. Finally, we design an ϵ-Nash learning algorithm, which adopts the θ-greedy exploration strategy, to derive the ϵ-approximate Nash equilibrium such that nodes can maximize their long-term payoffs. Our substantive simulation results and analysis verify that the proposed algorithm outperforms baseline algorithms in bringing higher payoffs to nodes and more fresh information to the control center.

Blood pressure (BP) is an essential vital sign related to many severe diseases, such as heart failure, kidney failure. Frequent BP detection can provide doctors more information to treat the disease. However, conventional at-home BP detection devices require completely blocking blood flow, which can lead to discomfort and disruption of normal activity when users want to perform frequent assessments. So a convenient solution should reduce the trouble of detecting BP in the daily life. In this work, we have designed and evaluated a smartphone-based BP detection application named WIB . WIB utilizes the smartphone’s acoustic sensors to obtain the chest motion caused by the heart beating and the smartphone’s camera to capture the pulse at the fingertip. We have recruited 30 volunteers who come from the author’s institution, to carry out comprehensive evaluations of WIB . We perform BP perturbation experiments to obtain different blood pressure data from these volunteers. The experiment results show that the average of Pearson correlation coefficient across all volunteers of the blood pressure estimation is 0.42–0.74 (α =0.6, σ =0.12), the average of RMSE across all volunteers is 4.2–8.8 mmHg (α =5.8, σ =1.8).

Speaker Recognition Systems (SRSs) grant access to legitimate users based on voiceprint. Recent research has shown that SRSs can be bypassed during the training phase (backdoor attacks) and the recognition phase (evasion attacks). In this paper, we explore a new attack surface of SRSs by presenting an enrollment-phase attack paradigm, named FenceSitter, where the adversary poisons the SRS using imperceptible adversarial ambient sound when the legitimate user registers into the SRS. The tainted voiceprint extracted by the SRS allows both the adversary and the legitimate user to access the system in all future recognition phases. To materialize such attack, we interleave carefully-designed continuous adversarial perturbations into innocent-sounding ambient sound. As computing adversarial perturbations over a long sequence of ambient sound carrier is intractable, we optimize over adversarial segments with content desensitization and physical realization. In addition, the attack is made available under the black-box settings by gradient estimation based on the natural evolution strategy. Extensive experiments have been conducted on both English and Chinese voice datasets for close-set identification (CSI), open-set identification (OSI), and speaker verification (SV) tasks. The results under various digital and physical conditions have verified the effectiveness and robustness of FenceSitter. With live enrollment experiments and user study, we further validate the practicality of FenceSitter. Our work reveals the vulnerability of SRSs during the enrollment phase, which may spur future research in improving the security of SRSs.

The blockchain with a distributed network structure can provide a reliable and secure environment for Internet of Things (IoT) transactions, which also suffers from low throughput, high computation overhead, and large transaction fee. Payment channel networks (PCNs) are developed to address the scalability issue of blockchain. A key enabler of PCNs is the path-finding services. Most of existing routing algorithms target at finding a single feasible path, which may lead to failure of large payments. Moreover, previous solutions did not consider the transaction fee of the chosen path, which is extremely important to cost-sensitive users. In this work, we design a new multipath routing algorithm for PCNs that aims at minimizing transaction fees. Together considering the determination of the optimal number of paths, the optimal path routes and the optimal allocation leads to difficulty of the problem. In addition, the transaction fees along the path are closely related to the amount of payment, and the capacity of a payment channel limits the payment that can be carried. To address these challenges, we propose MILPA-PCN, a cost-effective multipath routing framework for PCNs. We develop a genetic algorithm-based routing determination algorithm with carefully designed genetic operations. We evaluate MILPA-PCN based on the real trace of the lightning network, and verify that MILPA-PCN can reduce the transaction fee by 33.56% and improve the payment success rate by 14.45%.

The calculation of calorie consumption is of paramount importance for the human diet and health management. Most existing solutions use image-processing techniques to identify the food type and refer to the nutrition table to compute the total calorie, which is quite challenging to differentiate foods that look the same but contain vastly different quantities of calories. To address this issue, we propose to leverage near-infrared spectroscopy (NIRS) to derive the concentration of nutrients based on the unique absorption spectrum of foods. Instead of using the professional NIRS system that is bulky, expensive, and impractical for daily use by nonexpert users, we develop a low-cost portable NIRS system using commercial LEDs. As the quality of the signals of the low-power LEDs is relatively poor, we carefully design modulation schemes and interference elimination algorithms to improve the signal quality and remove the interference. Extensive experiments show that NIRSCAM outperforms the image-based baseline in achieving more accurate calorie estimation, especially for look-alike foods and is robust to various environmental factors.

Reputation/trust-based blockchain systems have attracted considerable research interests for better integrating Internet of Things with blockchain in terms of throughput, scalability, energy efficiency, and incentive aspects. However, most existing works only consider static adversaries. Hence, they are vulnerable to slowly adaptive attackers, who can target validators with high reputation value to severely degrade the system performance. Therefore, we introduce$\textsf{zkRep}$, a privacy-preserving scheme tailored for reputation-based blockchains. Our basic idea is to hide both the identity and reputation of the validators by periodically changing the identity and reputation commitments (i.e., aliases), which makes it much more difficult for slowly adaptive attackers to identify validators with high reputation value. To realize this idea, we utilize privacy-preserving Pedersen-commitment-based reputation updating and leader election schemes that operate on concealed reputations within an epoch. We also introduce a privacy-preserving identity update protocol that changes the identity and time-window-based cumulative reputation commitments during each epoch transition. We have implemented and evaluated$\textsf{zkRep}$on the Amazon Web Service. The experimental results and analysis show that$\textsf{zkRep}$achieves great privacy-preserving features against slowly adaptive attacks with little overhead.

Remote data integrity auditing ensures the integrity of cloud storage. In practice, cloud users may not want their sensitive data to be exposed to others. Thus, it is meaningful to investigate how to realize data sharing with sensitive information hiding in cloud storage auditing. Up to now, cloud storage has been proven to achieve the sensitive information hiding property through a third-party sanitizer dedicated to sanitize user data, which leads to high outlays on purchasing and maintaining a special server. To meet this challenge, we design a novel cloud storage auditing protocol to support sensitive information hiding without the need of a third-party sanitizer. In addition, our scheme allows data owners to enable or disable other users to access their sensitive information with the help of the cloud that dose not deviate from the agreement during access control. To be specific, only after receiving the delegations from the data owner, the users can compute the valid warrants that can pass the access verification of the cloud. The proposed protocol is built on identity-based cryptography, thus avoiding the complex certificate management. We validate the advantages of the proposed protocol through massive theoretical analysis and experimental results.

Fatigue is believed to be the leading factor for traffic accidents (e.g., fatigue driving) and health problems (e.g., heart disease and diabetes). However, fatigue-related risks are difficult to quantify because there is no efficient and reliable fatigue detection method comparable to blood alcohol testing for drunk drivers. Conventional fatigue detection methods either require wiring of sensors (e.g., EEG and ECG) that are inconvenient or leverage video camera systems that are lighting sensitive and may leak privacy. We present Ubi-Fatigue, a comfortable and contactless fatigue monitor system using wireless signals. Ubi-Fatigue combines both vital signs and facial features to achieve reliable fatigue detection. A series of novel signal recovery algorithms is developed to extract the heartbeat signal and the eye blink signal from the same raw signal captured by the single-antenna radar. We have implemented a fully functional prototype of Ubi-Fatigue using off-the-shelf radar. Twenty volunteers are involved in extensive experiments for a total duration of 480 h with more than 60 h of collected time-series data. The results demonstrate that Fatigue-Radio can reach a detection accuracy of 81.4%, which is higher than ECG- or visual-based fatigue detection systems, and is approximated to the ECG + visual-based fatigue detection system. Ubi-Fatigue is expected to provide a potential solution for smart home applications in the coming days.

With the prosperity of wireless and mobile communications, the allocation of wireless resources, e.g., spectrum channels, femtocell access permissions, and resource blocks of D2D connections, has become a matter of great concern, which leads to the emergence and popularity of electronic auction. However, the lack of privacy protection and fairness guarantee in the existing market has posed great obstacles to user participation in electronic auctions. Without privacy protection, users will be worried about the unauthorized exposure of their commercial secrets, which may hurt their payoffs in the long run. Without fairness guarantee, selfish/malicious users may prematurely abort the auction to avoid unsatisfactory payment, thus causing loss to honest users and wasting system resources. Furthermore, the lack of fairness guarantee will incur an exchange dilemma, where mutually distrusting parties may refuse to initiate the goods/money exchange, hindering the market-clearing of auctions. Although extensive efforts have been made to develop privacy-preserving auction mechanisms for wireless markets, the fairness issue, to the best of our knowledge, has never been taken into account in the context of auctions. Moreover, a truly practical framework that can be applied to most mainstream auction formats is also missing in the literature. In this article, we present the first general framework, named SAFE, for representative auctions in (but not limited to) wireless markets. SAFE achieves a wide range of security goals with much higher efficiency than existing solutions. Furthermore, SAFE ensures auction fairness in a trust-free manner by developing a full set of carefully-designed modular protocols that can be easily adapted to any auction format. We implement the SAFE framework over a simulated Ethereum network. Extensive experimental results confirm that SAFE can indeed achieve high economic efficiency in terms of social welfare and participation satisfaction at very low computation and communication overheads.

Cloud storage is more and more prevalent in practice, and thus how to check its integrity becomes increasingly essential. A classical solution is identity-based (ID-based) provable data possession (PDP), which supports certificateless cloud storage auditing without entire user data. However, existing ID-PDP protocols always require that cloud users outsource data blocks, authenticators and a small-sized file tag to the cloud, and make use of the heavy elliptic curve cryptography over bilinear pairing. These disadvantages would result in vast storage, communication, and computation costs, which is unexpected, especially for resource-limited cloud users. To improve the performance, this paper proposes a novel cryptographic primitive: ID-based PDP with compressed cloud storage. In this model, cloud storage auditing can be achieved by using only encrypted data blocks in a self-verified way, and original data blocks can be reconstructed from the outsourced data. Thus, data owners no longer need to store original data blocks on the cloud. We also use some basic algebraic operations to realize a concrete ID-based PDP protocol with compressed cloud storage, which is quite efficient due to no heavy cryptographic operations involved. The proposed protocol can easily be extended to support the other practical functions by using the primitive replacement technique. The proposed protocol is strictly proven to have the properties of correctness, privacy, unforgeability and detectability. Finally, we give plenty of theoretical analysis and experimental results to validate the efficiency of the proposed protocol.

Facial expressions are an essential form of human nonverbal communication. Recognition of this nonverbal sign may enable developers to understand the feedbacks on smart device functionality and advertising. Existing approaches for facial expression recognition are mainly based on cameras or on-body sensors, which are either sensitive to lighting conditions or cumbersome for users to wear devices on their faces. In this paper, we propose a new facial expression recognition system based on Wi-Fi signals, named WiFace. Our fundamental intuition is that facial muscle movements in different expressions will induce distinctive waveform patterns in the time-series of channel state information (CSI) in Wi-Fi signals. We develop a series of algorithms to process the CSI signals and extract the most representative waveform patterns for facial expression classification. We build a fully-functional prototype of WiFace using commercial off-the-shelf devices, which can recognize six typical facial expressions. We conduct extensive experiments to evaluate the performance of WiFace, and the experimental results show that the average recognition accuracy is 94.80 percent.

Myocardial infarction (MI) is the myocardial necrosis caused by persistent ischemia and hypoxia of coronary arteries. People do not realize that they are suffering from MI until they have a heart attack. Early MI detection plays a vital role in symptom relief and improvement in the performance of daily activities. However, conventional MI detection methods require expensive and inconvenient medical tests, e.g., intrusive blood tests or wear electrocardiogram (ECG) sensors, which can only be performed in medical institutions. In this paper, we introduce a contactless and non-intrusive MI detection method based on wireless sensing that monitors abnormalities in heartbeats. Specifically, we present Health-Radio, a radar-based system towards early MI detection. Health-Radio extracts heart rate variability (HRV) from the RF signals reflected from users. In particular, with our carefully designed signal processing algorithms, Health-Radio is able to not only obtain heartbeat signals when the user is stationary, but also tolerate interference when the user is performing certain activities, e.g., eating, reading and browsing the Internet. We have recruited 30 MI patients from the Central Hospital of Wuhan, China, and 30 healthy university students to conduct comprehensive evaluations of the performance of Health-Radio. The experiment results show that Health-Radio can achieve a median MI detection accuracy of 81.2 percent when the users are stationary, which is comparable to ECG-based MI detection. Even when the users are not stationary, Health-Radio can still achieve a median detection accuracy of 66.5 percent. Health-Radio is promising in providing a new paradigm for smart-home healthcare in the future.

Blockchain-based cryptocurrencies are severely limited in transaction throughput and latency. A promising solution to this issue is a payment channel, which allows trust-free payments between two peers without exhausting the resources of the blockchain. A linked payment channel network (PCN) enables payments between two peers through a series of intermediate nodes that forward and charge for the payments. However, most of existing proposals only use the shortest path as the path of the transaction, which causes the frequently reused channels to be exhausted quickly. In addition, most of existing PCNs are almost only designed for payments between two parties, which leads to limited application scenarios. When multiple payments use the same intermediate channel, the two-party PCNs cannot achieve simultaneous payments. In this paper, we propose a multi-party payment channel (MPC) network, a payment channel proposal that supports multiple payments using the same intermediate channel simultaneously, thereby greatly expanding the application scenarios of payment channels. In addition, our channel selection and transaction conversion strategies can also increase the success rate of transactions. We implement MPC network in the simulated blockchain network and lightning network based on Truffle, and a large number of experiments verify the effectiveness of our solution.

Blockchain-based cryptocurrencies are severely limited in transaction throughput and latency due to the need to seek consensus among all peers of the network. A promising solution to this issue is payment channels, which allow unlimited numbers of atomic and trust-free payments between two peers without exhausting the resources of the blockchain. A linked payment channel network enables payments between two peers without direct channels through a series of intermediate nodes that forward and charge for the transactions. However, the charging strategies of intermediate nodes vary with different payment channel networks. Existing works do not yet have a complete routing algorithm to provide the most economical path for users in a multi-charge payment channel network. In this work, we propose MPCN-RP, a general routing protocol for payment channel networks with multiple charges. Our extensive experimental results on both simulated and real payment channel networks show that MPCN-RP significantly outperforms the baseline algorithms in terms of time and fees.

As one of the biggest contributors to road accidents and fatalities, drink driving is worthy of significant research attention. However, most existing systems on detecting or preventing drink driving either require special hardware or require much effort from the user, making these systems inapplicable to continuous drink driving monitoring in a real driving environment. In this paper, we presentDetectDUI, a contactless, non-invasive, real-time system that yields a relatively highly accurate drink driving monitoring by combining vital signs (heart rate and respiration rate) extracted from in-car WiFi system and driver’s psychomotor coordination through steering wheel operations. The framework consists of a series of signal processing algorithms for extracting clean and informative vital signs and psychomotor coordination, and integrate the two data streams using a self-attention convolutional neural network (i.e., C-Attention). In safe laboratory experiments with 15 participants,DetectDUIachieves drink driving detection accuracy of 96.6% and BAC predictions with an average mean error of$2\sim 5mg/dl$. These promising results provide a highly encouraging case for continued development.

In the area of Internet of Things (IoT), voice assistants have become an important interface to operate smart speakers, smartphones, and even automobiles. To save power and protect user privacy, voice assistants send commands to the cloud only if a small set of preregistered wake-up words are detected. However, voice assistants are shown to be vulnerable to the FakeWake phenomena, whereby they are inadvertently triggered by innocent-sounding fuzzy words. In this paper, we present a systematic investigation of the FakeWake phenomena from three aspects. To start with, we design the first fuzzy word generator to automatically and efficiently produce fuzzy words instead of searching through a swarm of audio materials.We manage to generate 965 fuzzy words covering 8 most popular English and Chinese smart speakers. To explain the causes underlying the FakeWake phenomena, we construct an interpretable tree-based decision model, which reveals phonetic features that contribute to false acceptance of fuzzy words by wake-up word detectors. Finally, we propose remedies to mitigate the effect of FakeWake. The results show that the strengthened models are not only resilient to fuzzy words but also achieve better overall performance on original training datasets.

With recent advances in the study of biometrics, gait analysis has drawn much attention for its potential use in forensics, surveillance, and legal systems. In this paper, we present WIAGE, a contactless and non-intrusive gait-based age estimation system, which leverages wireless sensing to perform gait analysis to infer the age of individuals. Traditional age estimation systems either require users to carry wearable devices that are inconvenient or rely on image processing that is computationally intensive and sensitive to lighting conditions and occlusion. In contrast, WIAGE utilizes the incumbent WiFi infrastructure to infer the age of users with minimal interference to their activities. We adopt a series of signal processing techniques to recover clear gait patterns from the noisy WiFi signals and extract the most relevant features from steps that can be used for robust age estimation. The experimental results show that WIAGE can achieve an age estimation accuracy of 95.2% for 23 users, which demonstrates the feasibility and effectiveness of our proposed system.

Cloud service providers, including Google, Amazon, and Alibaba, have now launched machine-learning-as-a-service (MLaaS) platforms, allowing clients to access sophisticated cloud-based machine learning models via APIs. Unfortunately, however, the commercial value of these models makes them alluring targets for theft, and their strategic position as part of the IT infrastructure of many companies makes them an enticing springboard for conducting further adversarial attacks. In this paper, we put forth a novel and effective attack strategy, dubbed InverseNet, that steals the functionality of black-box cloud-based models with only a small number of queries. The crux of the innovation is that, unlike existing model extraction attacks that rely on public datasets or adversarial samples, InverseNet constructs inversed training samples to increase the similarity between the extracted substitute model and the victim model. Further, only a small number of data samples with high confidence scores (rather than an entire dataset) are used to reconstruct the inversed dataset, which substantially reduces the attack cost. Extensive experiments conducted on three simulated victim models and Alibaba Cloud's commercially-available API demonstrate that InverseNet yields a model with significantly greater functional similarity to the victim model than the current state-of-the-art attacks at a substantially lower query budget.

The advent of new network architectures has resulted in the rise of network applications with different network performance requirements: live video streaming applications require low latency. In contrast, file transfer applications require high throughput. Existing congestion control protocols may fail to simultaneously meet the performance requirements of these different types of applications since their designed objective function is fixed and difficult to readjust according to the needs of the application. In this paper, we develop MOCC (Multi-Objective Congestion Control), a novel multi-objective congestion control protocol that can meet the performance requirements of different applications without the need to redesign the objective function. MOCC leverages multi-objective reinforcement learning with preferences in order to adapt to different types of applications. By addressing challenges such as slow convergence speed and the difficulty of designing the end of the episode, MOCC can quickly converge to the equilibrium point and adapt multi-objective reinforcement learning to congestion control. Through an extensive array of experiments, we discover that MOCC outperforms the most recent state-of-the-art congestion control protocols and can achieve a trade-off between throughput, latency, and packet loss, meeting the performance requirements of different types of applications by setting preferences.

Traditional health insurance schemes make contracts with periodic premiums according to the health conditions of customers. Insurance companies are thus seeking an incentive for a healthy lifestyle of customers. Nowadays, more and more people want to gather metrics associated with physical activity. High quality and quantity information introduced by wearable devices may indicate the potential risks of acute conditions. Therefore, wearable technology brings new opportunities for insurance companies to change the interaction way with customers and to improve risk management for users. In this article, we investigate the problem of incentive mechanism design for the insurance market boosted by wearable technology. Designing an effective incentive mechanism is challenging since both the insurer and the users are selfish and rational parties who try to maximize their own utility. There exists an information asymmetry, i.e., the insurer is unaware of the intrinsic physical fitness of each user (type of users). Inspired by the contract theory, we design a series of optimal insurance contracts with different premium discounts and recommended exercise levels targeting different types of users. We first analyze the scenario where the insurer knows the type of each individual, then move to the more practical case where the insurer has incomplete information of user types. We theoretically prove that the proposed contracts satisfy individual rationality and incentive compatibility, which enables the insurer to achieve utility maximization and each type of user to choose the most appropriate contract. We have conducted extensive simulations to provide insight into the performance under different scenarios. The numerical results verify that our contract-based insurance incentive mechanism maximizes insurer's utility and motivates users to exercise more.

Crowdsourcing has become a popular paradigm to leverage the collective intelligence of massive crowd workers to perform certain tasks in a cost-effective way. Task assignment is an essential issue in crowdsourcing platforms owing to heterogeneous tasks and work skills. In this article, we focus on assigning workers with diversified skill levels to crowdsourcing tasks with different quality requirements and budget constraints. Task assignment is fundamentally a many-to-one matching problem, where one task is allocated to multiple users who can meet the minimum quality requirement of the task within the limited budget. While most existing works try to maximize the utility of the crowdsourcing platform, we take into account the individual preferences of crowdsourcers and workers toward each other to ensure the stability of task assignment results. In this article, we propose task assignment mechanisms that can guarantee stable outcomes for the many-to-one matching problem with lower and upper bounds (i.e., quality requirement and budget constraint) in regard to heterogeneous worker skill levels. Extensive simulation results show that the proposed algorithms can greatly improve the success ratio of task accomplishment and worker happiness compared with existing algorithms.

The time and monetary costs of training sophisticated deep neural networks are exorbitant, which motivates resource-limited users to outsource the training process to the cloud. Concerning that an untrustworthy cloud service provider may inject backdoors to the returned model, the user can leverage state-of-the-art defense strategies to examine the model. In this paper, we aim to develop robust backdoor attacks (named RobNet) that can evade existing defense strategies from the standpoint of malicious cloud providers. The key rationale is to diversify the triggers and strengthen the model structure so that the backdoor is hard to be detected or removed. To attain this objective, we refine the trigger generation algorithm by selecting the neuron(s) with large weights and activations and then computing the triggers via gradient descent to maximize the value of the selected neuron(s). In stark contrast to existing works that fix the trigger location, we design a multi-location patching method to make the model less sensitive to mild displacement of triggers in real attacks. Furthermore, we extend the attack space by proposing multi-trigger backdoor attacks that can misclassify inputs with different triggers into the same or different target label(s). We evaluate the performance of RobNet on MNIST, GTSRB, and CIFAR-10 datasets, against four representative defense strategies Pruning, NeuralCleanse, Strip, and ABS. The comparison with two state-of-the-art baselines BadNets and Hidden Backdoors demonstrates that RobNet achieves higher attack success rate and is more resistant to potential defenses.

Smart city communication networks hold the promise of harnessing the Internet of Things, smart devices, intelligent energy grids, autonomous cars, and many other data architectures to improve quality of life for every citizen. As time and progress have gradually shifted this concept from an idea into a reality, the possibilities for what services a smart city might provide have exploded. Along with that expansion, the number of devices that may need to be supported across a communications network has multiplied enormously. However, existing congestion control protocols are not equipped to support this higher requirement for network performance. New protocols are needed to provide higher data throughputs, reduce queuing delays and packet loss, and maintain stable and reliable communications pathways. The end-to-end congestion control protocol presented in this article does all these things. Called high throughput congestion control (HTCC), the framework takes full advantage of reinforcement learning and a fast growth algorithm to adaptively control the bytes in flight across the network's links to suit the prevailing network conditions. The result is a protocol that not only finds the optimal tradeoff between throughput, latency, and packet loss, but also significantly speeds up the learning process to avoid wasting network resources, use all free bandwidth and maximize network utilization. A comprehensive set of experiments with HTCC and six state-of-the-art protocols-Copa, PCC Vivace, PCC-Allegro, TCP Cubic, TaoVA-100x, and FillP-Sheep-demonstrate that, in most conditions, HTCC is able to make the best tradeoff between throughput, delay, and loss rate.

Vehicular ad hoc networks ( VANETs ) and the services they support are an essential part of intelligent transportation. Through physical technologies, applications, protocols, and standards, they help to ensure traffic moves efficiently and vehicles operate safely. This article surveys the current state of play in VANETs development. The summarized and classified include the key technologies critical to the field, the resource-management and safety applications needed for smooth operations, the communications and data transmission protocols that support networking, and the theoretical and environmental constructs underpinning research and development, such as graph neural networks and the Internet of Things. Additionally, we identify and discuss several challenges facing VANETs, including poor safety, poor reliability, non-uniform standards, and low intelligence levels. Finally, we touch on hot technologies and techniques, such as reinforcement learning and 5G communications, to provide an outlook for the future of intelligent transportation systems.

Influence maximization in social networks is of great importance for marketing new products. Signed social networks with both positive (friends) and negative (foes) relationships pose new challenges and opportunities, since the influence of negative relationships can be leveraged to promote information propagation. In this paper, we study the problem of influence maximization for advertisement recommendation in signed social networks. We propose a new framework to characterize the information propagation process in signed social networks, which models the dynamics of individuals' beliefs and attitudes towards the advertisement based on recommendations from both positive and negative neighbours. To achieve influence maximization in signed social networks, we design a novel Signed-PageRank (SPR) algorithm, which selects the initial seed nodes by jointly considering their positive and negative connections with the rest of the network. Our extensive experimental results confirm that our proposed SPR algorithm can effectively and efficiently influence a broader range of individuals in the signed social networks than benchmark algorithms on both synthetic and real datasets.

Following the correct stroke order while writing Chinese characters composed of strokes plays an important role in handwriting efficiency and quality, especially for early education. Most existing systems use image processing techniques for character and stroke order recognition, which is sensitive to lighting conditions. In this paper, we present the design, implementation and evaluation of SmartSO, which utilizes the inertial sensors of an off-the-shelf smartwatch for Chinese character and stroke order recognition. SmartSo first identifies the Chinese character written by the user, based on which SmartSo decides whether the stroke order is written correctly to help improve users' writing behavior. The biggest challenge for stroke order recognition is that some Chinese characters have repeated strokes (strokes of the same type), e.g., with two same horizontal strokes, and it is challenging to differentiate the writing order of such strokes given only the detected stroke composition (number and type of strokes). To mitigate this problem, we further analyze the hand movement between two adjacent strokes (referred to as direction motion) and propose a novel algorithm to recognize stroke order based on direction motion information. Finally, we build a fully functional prototype of SmartSO, and extensive experiments confirm its effectiveness and robustness.

Pattern lock is an essential authentication method on mobile devices. Recent works on cracking pattern locks either require additional network facilities (e.g., WiFi hotspots) or suffer from strict constraints (e.g., physical closeness to the victim and good lighting). Being too susceptible to environment settings, these attacks are less effective in practice and cannot scale to a large number of users. To address these concerns, in this paper, we propose PatternListener+, a practical attack on pattern locks using the speakers and microphones on mobile devices. The speaker plays inaudible acoustic signals, which are reflected by the fingertip when the victim is drawing the pattern, and then recorded by the microphone. The recorded acoustic signals contain rich information of the fingertip motion that can be leveraged to infer the pattern. We carefully design a series of algorithms to eliminate the dynamic and static interferences, segment acoustic signals into fragments corresponding to all pattern lines, and recover each line composed of the pattern according to the signals. Finally, we recover the candidate pattern by mapping all line candidates into grid patterns with a tree structure. We implement a PatternListener+ prototype using off-the-shelf smartphones, and extensive experiments confirm the effectiveness and robustness of PatternListener+. The attack success rate is over 90 percent on 120 patterns in five attempts.

With the widespread application of cloud storage, ensuring the integrity of user outsourced data catches more and more attention. To remotely check the integrity of cloud storage, plenty of protocols have been proposed, implemented by checking the equation constructed by the aggregated blocks, tags, and indices. However, the verifier only has the knowledge of the indices of the audited blocks and tags, which thus requires the cloud to store both data blocks and tags for integrity verification. In this article, we present a compressive secure cloud storage protocol inspired by Goldreich-Goldwasser-Halevi (GGH) cryptosystem. Since the aggregated blocks can be reconstructed from the aggregated tags without the help of data indices, the cloud can only store data tags for providing the verifiable integrity proof. In this way, communication and storage costs can be hugely reduced and user private information can be hidden from the cloud. Furthermore, the proposed protocol only contains a few basic algebraic operations, making it highly efficient. We also provide formal security proof of the proposed protocol regarding forge, replay and replace attacks. In addition, we explore a new technique to support data dynamics. Furthermore, we establish a generic framework of compressive secure cloud storage protocols. Finally, we provide the theoretical analysis and experimental results, which further validate the effectiveness of the proposed protocol.

Choosing the best cloud configuration for large-scale data analytics jobs deployed in the cloud can substantially improve their performance and reduce costs. However, current cloud providers offer a wide variety of instance types and customized cluster sizes, making it both time-consuming and costly to pinpoint the optimal cloud configuration. This article presents the design, implementation, and evaluation of Silhouette, a cloud configuration selection framework based on performance models for various large-scale analytics jobs with minimal training overhead. The essence of Silhouette is to build performance prediction models with carefully selected small-scale experiments on small subsets of input data to estimate the performance with entire input data on larger cluster sizes. To reduce the training time and cost, Silhouette incorporates new statistical techniques to select those experiments that yield the best possible information for performance prediction. Moreover, we develop a novel model transformer to convert a prediction model built on one instance type to a different instance type with only one extra experiment, which significantly reduces the training overhead. We evaluate Silhouette with an extensive array of large-scale data analytics jobs on Amazon EC2. Our experimental results have shown convincing evidence that Silhouette is effective in optimizing cloud configuration while saving both training time and costs compared with existing solutions.

Traditional congestion control algorithms were designed with a hardwired heuristic mapping between packet- level events and predefined control actions in response to these events, and may fail to satisfy all the desirable performance goals as a result. In this paper, we seek to reconsider these fundamental goals in congestion control, and propose Eagle, a new congestion control algorithm to refine existing heuristics. Eagle takes advantage of expert knowledge from an existing algorithm, and uses deep reinforcement learning (DRL) to train a generalized model with the hope of learning from an expert. Learning by trial-and-error may not be as efficient as imitating a teacher; by the same token, DRL alone is not enough to guarantee good performance. In Eagle, we seek help from an expert congestion control algorithm, BBR, to help us train a long-short term memory (LSTM) neural network in the DRL agent, with the hope of making decisions that can be as good as or even better than the expert. With an extensive array of experiments, we discovered that Eagle is able to match and even outperform the performance of its teacher, and outperformed a large number of recent congestion control algorithms by a considerable margin.

With the explosion of the Internet-of-Things (IoT) technology, numerous IoT terminal devices generate tremendous traffic. Device-to-device (D2D)-enabled caching can greatly relieve the pressure of massive resource-limited terminal devices in the IoT network. This article proposes a novel distributed framework, termed crowdcaching, which motivates selective file caching and cooperative file sharing among terminal devices via short-range (e.g., D2D) communications. After modeling file preference distributions and local connectivities, we optimize the caching strategy for any given coalition of cooperative users to minimize their total delay cost. In particular, if users have homogeneous file preferences and local connectivities, we can mathematically define a popularity index, according to which files are chosen to be cached in user devices. In a more general setting where users have heterogeneous file preferences and local connectivities, we propose a greedy algorithm of low complexity to determine the optimal caching strategy. Based on the cooperative caching strategy for any given coalition, we further investigate users' incentive to form crowdcaching coalitions through the coalitional game theory and propose a distributed algorithm to yield a stable coalition formulation. The simulation results show that crowdcaching can effectively reduce the average delay cost of users by as much as 45.64%.

The explosive growth of the Internet of Things (IoT) is posing a high pressure on the spectrum resources. Dynamic spectrum allocation is an effective solution for supporting massive IoT devices with limited spectrum resources, often implemented in the form of spectrum auction. Among diverse auction formats, double auctions feature multiple buyers and multiple sellers, which is applicable to a wide range of IoT scenarios. Nevertheless, the existing double auction mechanisms mostly focus on the design of truthfulness and ignore the importance of privacy preservation. Hence, in this article, we propose PP-SPEC, a truthful and privacy-preserving double auction mechanism that takes spatial reuse, spectrum heterogeneity, and multiminded bidders into account, and provides full privacy in terms of asking prices, bidding values, and locations. To tackle the challenge of privacy-preserving comparison over ciphertext field, we carefully designed a set of generic privacy-preserving ciphertext comparison protocols (PIC), including basic PIC (B-PIC) and extended PIC (E-PIC), by integrating additive homomorphic encryption and garbled circuits. Furthermore, we have theoretically proved the truthfulness and security of PP-SPEC. We have implemented and evaluated PP-SPEC, whose experimental results show high performance and practicability of PP-SPEC for real-life applications.

Most existing systems use portable devices or image processing techniques for handwritten Chinese character recognition (HCCR), which are unable to detect character when writing on a paper or sensitive to lighting conditions. In this article, we present the design, implementation, and evaluation of a smartwatch-based HCCR system, called SmartHandwriting. To segment each Chinese character, we further analyze the hand movement between the handwriting gesture and the wrist movement gesture and propose a novel algorithm to distinguish the two types of gestures. Due to too many Chinese characters for classification, we utilize the data augmentation method for avoiding overfitting. Then, we build the HCCR model using the deep convolutional neural network (DCNN) method. The recognition accuracy of the Chinese characters is 96.0%, and extensive experiments confirm its effectiveness and robustness. Moreover, we also explore adverse factors that affect the recognition performance, which can be avoided in the future.

Voice authentication is becoming increasingly popular, which offers potential benefits over knowledge and possession based authentication methods. Meanwhile, the unique features of lip movements during speaking have been proved to be useful for authentication. However, the unimodal biometric authentication systems based on either voice or lip movements have certain limitations. Voice authentication systems are prone to spoofing attacks and suffer from serious performance degradation in noisy environments. Lip movements authentication systems are unstable and are sensitive to the user's physical and psychological conditions. In this paper, we propose and implement LVID, a multimodal biometrics authentication system on smartphones, which resolves the defects of the original systems by combining the advantages of lip movements and voice. LVID simultaneously captures these two biometrics with the built-in audio devices on smartphones and fuses them at the data level. The reliable and effective features are then extracted from the fused data for authentication. LVID is practical as it requires neither cumbersome operations nor additional hardwares but only a speaker and a microphone that are commonly available on smartphones. Our experimental results with 104 participants show that LVID can achieve 95% accuracy for user authentication, and 93.47% of the attacks can be detected. It is also verified that LVID works well with different smartphones and is robust to different smartphone positions.

With powerful parallel computing GPUs and massive user data, neural-network-based deep learning can well exert its strong power in problem modeling and solving, and has archived great success in many applications such as image classification, speech recognition and machine translation etc. While deep learning has been increasingly popular, the problem of privacy leakage becomes more and more urgent. Given the fact that the training data may contain highly sensitive information, e.g., personal medical records, directly sharing them among the users (i.e., participants) or centrally storing them in one single location may pose a considerable threat to user privacy. In this paper, we present a practical privacy-preserving collaborative deep learning system that allows users to cooperatively build a collective deep learning model with data of all participants, without direct data sharing and central data storage. In our system, each participant trains a local model with their own data and only shares model parameters with the others. To further avoid potential privacy leakage from sharing model parameters, we use functional mechanism to perturb the objective function of the neural network in the training process to achieve ε-differential privacy. In particular, for the first time, we consider the existence of unreliable participants, i.e., the participants with low-quality data, and propose a solution to reduce the impact of these participants while protecting their privacy. We evaluate the performance of our system on two well-known real-world datasets for regression and classification tasks. The results demonstrate that the proposed system is robust against unreliable participants, and achieves high accuracy close to the model trained in a traditional centralized manner while ensuring rigorous privacy protection.

The dramatic growth of mobile application usage has posed great pressure on application developers to better manage their backend capacity. Rule-based or schedule-based auto-scaling mechanisms have been proposed, but it is difficult or expensive to frequently adjust the backend capacity to track the burstiness of mobile traffic. In this paper, we explore a fundamentally different approach. Instead of scaling the backend in line with the mobile traffic, we smooth out traffic profiles to reduce the required backend capacity and increase its utilization. Our proposed solution, called Razor, is inspired by two key insights on mobile traffic. First, mobile traffic exhibits high short-term fluctuations but steady long-term trend, so that we may temporarily delay user requests and periodically adapt backend capacity based on the predicted traffic volume. Second, user requests have different priorities: while some requests are urgent (e.g., sending a message), some are delay-tolerant (e.g., changing the profile photo) and can be postponed without much influence on the user experience. Based on these observations, our design features a two-tier architecture: on a long timescale, Razor predicts future traffic using machine learning algorithms and plans the optimal backend capacity to minimize the budget with performance guarantee; on a short timescale, Razor schedules which requests to delay and by how much time to delay according to their delay tolerance. We implement a fully-functional prototype of Razor, and evaluate its performance with both real and synthetic traces. Extensive experimental results show that Razor can effectively help mobile application developers reduce their backend cost while guaranteeing the user experience.

Mining pools have become dominant in today's bitcoin mining network, where miners can pool their powers together for reduced variance of block mining and steadier stream of potential income. Along with the continuous evolvement of mining pools are the increasingly intense competitions among them. Recent empirical studies have shown that the distributed denial-of-service (DDoS) attack is one of the most common ways for competing mining pools to sabotage the rivals and earn illegitimate rewards. Existing efforts have been made on using static game models to analyze the interactions between mining pools, and derive the Nash Equilibrium and optimal attacking strategies in a one-time static context. To better understand the impact of such DDoS attacks, in this paper, we take a starkly different approach, and for the first time address the dynamics in mining pool attacks. Specifically, we start by formulating the interactive competition among mining pools as a general-sum stochastic game. Then we propose an efficient Nash learning algorithm to obtain the near optimal attacking strategy that maximizes the expected long-term utility. Our theoretical analysis and extensive experimental results both show that the proposed strategy outperforms the baseline myopic learning algorithm, which only aims at maximizing the revenue in the current time stage. These findings, together with our proposed stochastic game model and learning algorithm, are expected to provide more practical guidelines for mining pools to survive and thrive in the highly-competitive bitcoin ecosystem.

The potential of multiuser MIMO (MU-MIMO) to increase network capacity has been intensively studied. To enable concurrent transmission in Frequency-Division Duplex (FDD) systems with limited feedback, users have to report the estimated channel state information (CSI) to the base station (BS) for interference elimination, which however has been proven to be vulnerable. In this paper, we reveal how to utilize the CSI feedbacks to launch sniffing attacks in a deterministic way, where thesniffing attackenables the attackers to eavesdrop other users in the same transmission group. We conduct a rigorous theoretical analysis on the construction of the forged CSI. In the proposed sniffing attacks, the malicious users can successfully sniff other users’ messages without knowing their own messages for signal cancellation. To make sniffing attacks more practical, we explore the possibilities of sniffing multiple users by a coalition of malicious users or a single malicious user. To thwart sniffing attacks, we design a novel secure CSI feedback (SCF) protocol based on the random masking technique, which requires simple modifications at the BS and incurs a little additional workload. Extensive theoretical analysis and experimental results are provided to further validate the effectiveness of our proposed sniffing attacks and the corresponding countermeasure.

Nowadays, smartphones have become indispensable for people and many users may browse information on their smartphones during walking. This raises severe concerns about pedestrian safety when distracted users inadvertently walks from the sidewalk onto the street. In this paper, we present WatchOut, a convenient pedestrian alerting system based on fine-grained step classification (e.g., flat, up and down the ramp) and event detection (e.g., entering the street, turn) in an urban environment. In contrast to existing systems that rely on shoe-mounted sensors, Watchout leverages rich sensors on the smartphone without requiring any additional hardwares. With carefully designed realtime data processing and classification algorithms, WatchOut can achieve a high detection accuracy. We develop a fully-functional Android application of WatchOut with user-friendly interface that will issue timely alerts to warn users of unsafe conditions. Extensive experiments with 23 volunteers in real pedestrian environment have confirmed the effectiveness of WatchOut, which achieves a comparable detection accuracy as the shoe-mounted system with off-the-shelf smartphones.

Social tie prediction is an important issue in social network analysis. Transfer learning is often used for social tie prediction to address the problem of insufficient labeled training data, since few users manually annotate their social relationships. In this paper, we propose TranGAN, a novel generative adversarial network (GAN) based transfer learning framework for social tie prediction, which leverages social theories as the common knowledge to bridge the source network and the target network. GAN helps augment the original data set by generating data samples that have a similar probability distribution to that of the original data, and the training of TranGAN converges faster compared to existing transfer learning models. We evaluate the performance of TranGAN with extensive experiments, and show that TranGAN outperforms traditional learning algorithms and existing transfer learning algorithm on several metrics, and is efficient for large-scale social networks.

Performing the correct pen-holding gesture plays an important role in handwriting efficiency and quality, especially for early education. In this paper, we present the design, implementation and evaluation of SmartWriting, which utilizes the smartwatch for pen-holding gesture recognition for both Chinese and English writing. SmartWriting can automatically identify whether the user is writing Chinese or English, based on which two classifiers are built to detect pen-holding gestures by strokes in Chinese characters or English letters. In particular, we propose to leverage the combined signal of one vertical stroke and one horizontal stroke for efficient pen-holding gesture identification in Chinese writing, and we are able to infer pen-holding gesture according to any letter in English writing. We build a fully functional prototype of SmartWriting, and extensive experiments confirm its effectiveness and robustness. The detection accuracy is 9. % and 9 % for Chinese and English respectively. SmartWriting provides a natural, convenient and inexpensive way to improve users' writing habits.

Voice biometrics is widely adopted for identity authentication in mobile devices. However, voice authentication is vulnerable to spoofing attacks, where an adversary may deceive the voice authentication system with pre-recorded or synthesized samples from the legitimate user or by impersonating the speaking style of the targeted user. In this paper, we design and implement VoicePop, a robust software-only anti-spoofing system on smartphones. VoicePop leverages the pop noise, which is produced by the user breathing while speaking close to the microphone. The pop noise is delicate and subject to user diversity, making it hard to record by replay attacks beyond a certain distance and to imitate precisely by impersonators. We design a novel pop noise detection scheme to pinpoint pop noises at the phonemic level, based on which we establish individually unique relationship between phonemes and pop noises to identify legitimate users and defend against spoofing attacks. Our experimental results with 18 participants and three types of smartphones show that VoicePop achieves over 93.5% detection accuracy at around 5.4% equal error rate. VoicePop requires no additional hardware but only the built-in microphones in virtually all smartphones, which can be readily integrated in existing voice authentication systems for mobile devices.

Spectrum auction is an effective way to redistribute scarce spectrum resources. However, most spectrum auction designs only target at economic robustness, while neglecting the inherent privacy leakage problem. Existing secure spectrum auction mechanisms fail to provide adequate security, and they all neglect the online fashion of spectrum request arrival. In this paper, for the first time, we propose a Privacy-pReserving and truthful Online double auction mechanism for Spectrum allocaTion in wireless networks, PROST. Compared with the state-of-the-art solutions, PROST provides a comprehensive and strong protection for users' sensitive information, especially for location privacy and time dynamics. PROST is constructed based on our carefully designed security building blocks, which support various arithmetics over encrypted real numbers, and they are also well applicable in other spectrum auctions. Besides, we improve on the existing online spectrum auction mechanisms by designing a novel privacy-preserving buyer grouping protocol for spectrum reuse. We not only theoretically prove that PROST can realize an all-round security against semi-honest adversaries but also extensively evaluate its performance. Experimental results validate that PROST achieves nice spectrum allocation efficiency with light computation and communication costs.

Auction is an effective way to allocate goods or services to bidders who value them the most. The rapid growth of e-auctions facilitates online transactions but poses new and distinctive challenges. It is difficult to establish trust among sellers, buyers, and auctioneers without centralized auction websites or platforms (the auctioneer) which collect bids and derive the auction results. However, these third parties may be untrustworthy, and malicious sellers or buyers may refuse to deliver the goods or payment according to the protocol. Moreover, the open and anonymous online environment may stimulate auction participants to form collusion coalitions to rig the auction and reap unfair profit. Many auction designs have been proposed to address these concerns, but they fall short of simultaneously achieving decentralization (i.e., held without a trusted third utility), strong consensus (i.e., the establishment of trust), collusion resistance, and practical implementation. We present CReam, the first decentralized collusion-resistant e-auction system that is implemented with smart contract on the blockchain. With the carefully designed smart auction contract, mutually distrustful and rational sellers and buyers are stimulated to operate properly, hence transact safely without trusted third parties. The auction mechanism in the smart contract can effectively prevent bidder collusion and realize economic robustness, i.e., truthfulness. We implement a fully functional CReam on the Ethereum network. Extensive experimental results confirm that CReam can greatly reduce the probability of collusion and achieve an approximate optimal revenue at a low cost of contract execution.

Dynamic spectrum allocation via auction is an effective solution to spectrum shortage. Combinatorial spectrum auction enables buyers to express diversified preferences towards different combinations of channels. Despite the effort to ensure truthfulness and maximize social welfare, spectrum auction also faces potential security risks. The leakage of sensitive information such as true valuation and location of bidders may incur severe economic damage. However, there is a lack of works that can provide sufficient protection against such security risks in combinatorial spectrum auction. In this paper, we propose ARMOR, to enable combinatorial auction for heterogeneous spectrum with privacy, which can preserve bidders' privacy while guaranteeing the economic-robustness of the combinatorial auction. We leverage the cryptographic methods, including homomorphic encryption, order-preserving encryption, and garbled circuits, to shield the bid and location information of buyers from the auctioneer. We design a novel location protection algorithm, which allows the auctioneer to exploit spectrum reuse opportunities without knowing the exact locations of buyers. Furthermore, we propose a verifiable payment scheme based on digital signature to prevent the auctioneer from forging the payment. The extensive experiments confirm that ARMOR maintains the good performance of the combinatorial spectrum auction, in terms of buyer satisfactory ratio and social welfare, and achieves privacy preservation with acceptable computation and communication costs.

Dual-channel screen-camera communication has been proposed to enable simultaneous screen viewing and hidden screen-camera communication. However, it strictly requires a well-controlled camera-screen alignment and an obstacle-free access. In this paper, we propose Dolphin, a novel real-time acoustics-based dual-channel communication system. Leveraging masking effects of human auditory system and readily available audio signals, Dolphin enables real-time unobtrusive speaker-microphone data communication without affecting the primary audio-hearing experience of human users. Compared with screen-camera communication, Dolphin supports non-line-of-sight transmissions and more flexible speaker-microphone alignments. Dolphin can also automatically adapt the data rate to various channel conditions. We further develop a secure data broadcasting scheme on Dolphin, where only designated privileged users can recover the embedded information in the acoustic signals. Our Dolphin prototype, built using COTS (Commercial Off-The-Shelf) smartphones, realizes (potentially secure) real-time hidden information communication, supports up to 8-meter signal capture distance and +900 listening angle, and achieves an average goodput of 240 bps at 2 m.

In a crowdsourcing system, it is important for the crowdsourcer to engineer extrinsic rewards to incentivize the participants. With mobile social networking, a user enjoys an intrinsic benefit when she aligns her behavior with the behavior of others. Referred to as network effects , such an intrinsic benefit becomes more significant as more users join and contribute to the crowdsourcing system. But should a crowdsourcer design her extrinsic rewards differently when such network effects are taken into consideration? In this article, we incorporate network effects as a contributing factor to intrinsic rewards, and study its influence on the design of extrinsic rewards. We show that the number of participating users and their contributions to the crowdsourcing system evolve to a steady equilibrium, thanks to subtle interactions between intrinsic rewards due to network effects and extrinsic rewards offered by the crowdsourcer. Taken network effects into consideration, we design progressively more sophisticated extrinsic reward mechanisms, and propose new and optimal strategies for a crowdsourcer to obtain a higher utility. Through simulations and examples, we demonstrate that with our new strategies, a crowdsourcer is able to attract more participants with higher contributed efforts; and the participants gain higher utilities from both intrinsic and extrinsic rewards.

Over the past decades, there have been extensive research endeavors in spectrum auction design. However, most solutions only focus on the allocation efficiency while ignoring the privacy leakage inherent in the process of spectrum auction. So far, the very few existing works on secure spectrum auctions either provide inadequate privacy protection or incur performance loss in terms of spectrum reusability. In this paper, for the first time, we propose PS-TAHES, a privacy-preserving and truthful double auction mechanism for heterogeneous spectrum. PS-TAHES is constructed based on our carefully designed security primitives, which can support various arithmetics over encrypted data, including multiplication, bid comparison, and sorting matrix, and they are well applicable in other contexts. We theoretically analyze the security and efficiency of PS-TAHES, which is proved to ensure a full and strong privacy protection for bidders while preserving the allocation efficiency of the original auction mechanism. Experimental results, consistent with the theoretical analysis, further validate the practical use of PS-TAHES in real-world applications.

IPTV provides a variety of services to users, e.g., Live TV, Video on Demand (VOD), and Catch-up TV, which allows users to freely select from a huge pool of program genres. A high-quality program recommendation system that can well predict users' dynamic preferences is desirable to improve user satisfaction. In this paper, we make the first attempt to design a personalized TV program recommendation system based on tensor learning that leverages temporal context information and implicit feedback. We establish the user preference model by carefully analyzing the influence of program genres and time context on users' viewing behavior in IPTV viewing logs. To obtain the optimal program recommendation, we adopt tensor decomposition to mine the latent relationship among users, program genres and the time contexts, while address issues of data sparsity and missing information. Our extensive experimental results on a real dataset from a major IPTV service provider in China have confirmed the superiority of our proposed recommendation algorithm over baseline algorithms in achieving a higher recommendation accuracy and covering a wider range of programs.

The use of a combinatorial auction is believed to be an effective way to distribute spectrum to buyers who have diversified valuations for different spectrum combinations. However, the allocation of spectrum with combinatorial auctions mainly aims at optimizing over certain utility functions, e.g., social welfare, but ignores individual preferences of buyers and sellers, who have incentives to deviate from globally optimal allocation results to improve their own utility. In this paper, we explore the possibility of designing a new stable matching algorithm for combinatorial spectrum allocations. Starkly different from existing efforts on spectrum matching mechanism design, our proposed combinatorial spectrum matching framework not only allows buyers to express preferences towards spectrum combinations (rather than individual channels), but also computes the payment that should be transferred from buyers to sellers. Payment determination, while essential in spectrum exchange, has never been addressed in existing spectrum matching frameworks. We design a novel algorithm to achieve a stable combinatorial spectrum matching and to compute the corresponding payment profiles. We conducted an extensive array of experiments to compare the performance of stable combinatorial spectrum matching with spectrum auctions. It is shown that the combinatorial spectrum matching sacrifices little allocation efficiency in terms of social welfare and spectrum utilization, but achieves a much higher individual buyer utility, which will incentivize buyers to participate and comply with the allocation results.

Image retrieval is crucial for social media sites such as Instagram to identify similar images and make recommendations for users who share similar interests. To get rid of the storage burden and computation for image retrieval, outsourcing to a remote cloud is now a trend. Yet, privacy concerns mandate the use of encryption before outsourcing the images. We need a secure way for retrieving images from a not-fully-trusted server. This paper proposes InstantCryptoGram, a secure image retrieval service. We first design a new data structure called sub-simhash, which fits for the inverted index used by many searchable symmetric encryption schemes. It leads to our modular solution that supports efficient similarity queries and updates over encrypted images. Our experiments on Amazon AWS EC2 over representative datasets show that our scheme is efficient and accurate in finding similar images while preserving privacy.

Data mining has heralded the major breakthrough in data analysis, serving as a “super cruncher” to discover hidden information and valuable knowledge in big data systems. For many applications, the collection of big data usually involves various parties who are interested in pooling their private data sets together to jointly train machine-learning models that yield more accurate prediction results. However, data owners may not be willing to disclose their own data due to privacy concerns, making it imperative to provide privacy guarantee in collaborative data mining over distributed data sets. In this paper, we focus on tree-based data mining. To begin with, we design novel privacy-preserving schemes for two most common tasks: regression and binary classification, where individual data owners can perform training locally in a differentially private manner. Then, for the first time, we design and implement a privacy-preserving system for gradient boosting decision tree (GBDT), where different regression trees trained by multiple data owners can be securely aggregated into an ensemble. We conduct extensive experiments to evaluate the performance of our system on multiple real-world data sets. The results demonstrate that our system can provide a strong privacy protection for individual data owners while maintaining the prediction accuracy of the original trained model.

Nowadays, machine learning is becoming a new paradigm for mining hidden knowledge in big data. The collection and manipulation of big data not only create considerable values, but also raise serious privacy concerns. To protect the huge amount of potentially sensitive data, a straightforward approach is to encrypt data with specialized cryptographic tools. However, it is challenging to utilize or operate on encrypted data, especially to perform machine learning algorithms. In this paper, we investigate the problem of training high quality word vectors over large-scale encrypted data (from distributed data owners) with the privacy-preserving collaborative neural network learning algorithms. We leverage and also design a suite of arithmetic primitives (e.g., multiplication, fixed-point representation, sigmoid function computation, etc.) on encrypted data, served as components of our construction. We theoretically analyze the security and efficiency of our proposed construction, and conduct extensive experiments on representative real-world datasets to verify its practicality and effectiveness.

To enable dynamic spectrum access, service providers with spare spectrum (sellers) trade with those who are in need of additional spectrum (buyers). In a spectrum market, the transaction result is essentially a matching between sellers and buyers. Though it is tempting to optimize the matching over certain utility functions, a stable matching is more desirable, since no participants have incentives to deviate from the matching result. Existing spectrum matching algorithms only consider the maximum number of channels a buyer can purchase, but ignore minimum spectrum requirement that is essential to support proper operation of wireless communications. In this paper, we present a new framework of spectrum matching with both maximum quota and minimum requirements. Different from conventional matching problems, the spectrum market poses distinctive challenges due to spectrum reusability. To tackle this problem, we design two novel algorithms that satisfy different stability criterion: Extended Deferred Acceptance (EDA) algorithm that is fair but wasteful and the Multistage Deferred Acceptance (MDA) algorithm that is non-wasteful but weakly fair. Both algorithms converge to an interference-free matching and guarantees the minimum spectrum requirement. The simulation results show that the two proposed algorithms can raise buyer happiness and the channel utilization.

In crowdsourcing systems, job assignment is one of the fundamental concerns. Existing works address the job assignment problem solely from the perspective of the crowdsourcer, aiming at maximizing the utility or minimizing the cost for the crowdsourcer. In this paper, we take into consideration users' preferences towards different jobs, and propose a novel matching framework for job assignment in crowdsourcing systems. Assigning multiple users to the same job will improve its quality, however, the crowdsourcer has to guarantee that the total payment to these users is less than the budget of this job. As users ask for different payments due to heterogeneity in their quality levels, classic deferred acceptance algorithm cannot reach a stable job assignment. In this paper, we formulate the job assignment problem in crowdsourcing systems as a many-to-one matching with budget constraints. Then, we design an algorithm that produces a stable job assignment in spite of user heterogeneity. Simulation results show that the proposed job assignment algorithm yields a high average job quality with a low computational complexity.

Mobile crowdsensing (MCS) is emerging as a new paradigm to perform crowdsourcing tasks. While incentive mechanisms for mobile users in the MCS market are well studied, there is a lack of study on the competition of crowdsourcers for limited sensing resources and data processing capacity. In this paper, we consider a MCS market, where multiple crowdsourcers bid for sensing resources provided by an MCS platform to perform crowdsourcing tasks. The MCS platform allocates sensing resources to selected crowdsensing tasks and charge crowdsourcers corresponding prices. To begin with, we formulate the sensing resource allocation as a social welfare maximization problem, and propose a primal-dual approximation algorithm that achieves a feasible solution in polynomial time with guaranteed approximation ratio. Then, we design a payment scheme to ensure the truthfulness and individual rationality of the auction. Both theoretical analysis and simulation results verify the efficiency and effectiveness of our proposed auction mechanism for resource provisioning in the MCS market.

With the development of cloud storage, data owners no longer physically possess their data and thus how to ensure the integrity of their outsourced data becomes a challenging task. Several protocols have been proposed to audit cloud storage, all of which rely mainly on data block tags to check data integrity. However, their block tag constructions employ cryptographic operations, which makes them computationally complex. In this paper, we investigate a secure cloud storage protocol based on the classic discrete logarithm problem. Our protocol generates data block tags with only basic algebraic operations, which brings substantial computation savings compared with previous work. We also strictly prove that the proposed protocol is secure under a definition which captures the real-world uses of cloud storage. In order to fit more application scenarios, we extend the proposed protocol to support data dynamics by employing an index vector and third-party public auditing by using a random masking number, both of which are efficient and provably secure. At last, theoretical analysis and experimental evaluation are provided to validate the superiority of the proposed protocol.

Crowdsourcing leverages the collective intelligence of the massive crowd workers to accomplish tasks in a cost-effective way. On a crowdsourcing platform, it is challenging to assign tasks to workers in an appropriate way due to heterogeneity in both tasks and workers. In this paper, we explore the problem of assigning workers with various skill levels to tasks with different quality requirements and budget constraints. We first formulate the task assignment as a many-to-one matching problem, in which multiple workers are assigned to a task, and the task can be successfully completed only if a minimum quality requirement can be satisfied within its limited budget. Different from traditional task assignment mechanisms which focus on utility maximization for the crowdsourcing platform, our proposed matching framework takes into consideration the preferences of individual crowdsourcers and workers towards each other. We design a novel algorithm that can generate a stable outcome for the many-to-one matching problem with lower and upper bounds (i.e., quality requirement and budget constraint), as well as heterogeneous worker skill levels. Through extensive simulations, we show that the proposed algorithm can greatly improve the success ratio of task accomplishment and worker happiness, when compared with existing algorithms.

To enable dynamic spectrum access, service providers with spare spectrum (sellers) trade with those who are in need of additional spectrum (buyers). In a spectrum market, the transaction result is essentially a match between sellers and buyers. Though it is tempting to optimize the matching over certain utility functions, a stable matching is more desirable, since it takes into account a diverse set of preferences of buyers and sellers, and produces a matching result which no participants have incentives to deviate from. While existing works on spectrum matching only consider the maximum number of channels a buyer can purchase, in real-world scenarios, the minimum spectrum requirement should be satisfied to support the proper operation of wireless communication. To address this issue, in this paper, we present a new framework of spectrum matching with both maximum and minimum requirements. Different from conventional matching problems, the spectrum market poses distinctive challenges due to spectrum reusability. Instead of being sold exclusively to just one buyer, the same channel can be reused by multiple buyers who are not interfering with each other. To tackle this problem, we design a novel algorithm, called Extended Deferred Acceptance (EDA), that converges to an interference-free matching and guarantees the minimum spectrum requirement. We theoretically prove the stability of the matching result. Our simulation results show that EDA can achieve a 100% coverage on the minimum requirements, while alternative benchmark algorithms fail to do so, and buyers are more satisfied with the matching result of EDA than that of alternative algorithms.

It is essential for mobile application developers to manage backend resources to serve dynamic user requests from the frontend. For a typical mobile application, the rate at which the user requests arrive at the backend fluctuates dramatically. However, it is difficult or expensive to frequently adjust the capacity of the backend to meet the request demand. In this paper, we present Clockwork, a third-party cloud service, which smooths the demand profile by redistributing delay-tolerant requests and prioritizing delay-sensitive requests, so that sufficient capacity can be provided with reduced cost and wastage. To begin with, Clockwork plans the optimal backend capacity on a relatively long timescale based on future demand estimated by machine learning algorithms. We discuss pros and cons of various simple machine learning algorithms and advanced deep learning algorithms, in terms of their prediction accuracy and training time. Then, Clockwork schedules user requests on a shorter timescale through a fair and Pareto- optimal rate allocation. We implemented a fully-functional prototype of Clockwork on cloud servers and user mobile devices. The experimental results show that Clockwork can effectively help developers cut cost, as well as improve the backend utilization.

Dynamic spectrum access (DAS) is an efficient way to redistribute spare channels among users. Conventionally, dynamic spectrum access is conducted through (double) spectrum auction, where a third-party auctioneer collects bids from buyers and sellers, and determines the spectrum allocation. Rather than placing bids only on individual channels, combinatorial spectrum auction allows buyers to express their valuations for different combinations of channels. However, auction mechanisms are generally vulnerable to the collusion between the auctioneer and buyers or sellers. Furthermore, to find the optimal allocation in combinatorial auction is usually NP-hard. In this paper, we propose to leverage a many-to-many matching framework to realize combinatorial spectrum trading. Unlike traditional many-to-many matching problem, spectrum matching is more challenging, because spectrum allocation is interference-limited rather than quota-limited. To deal with this problem, we propose a novel matching algorithm, which takes buyers' interference relationship into consideration. We theoretically prove that the matching result is individual rational, strong pairwise stable and is a subgame-perfect Nash equilibrium of the corresponding spectrum bargaining game. Simulation results show that the proposed algorithm can converge to a stable matching within a few iterations.

Dynamic spectrum access (DSA) redistributes spectrum from service providers with spare channels to those in need for them. Existing works on such spectrum exchange mainly focus on double auctions, where an auctioneer centrally enforces a certain spectrum allocation policy. In this paper, we take a different and new perspective, proposing to use matching as an alternative tool to realize DSA in a distributed way for a free market, which consists of only buyers and sellers, but no trustworthy third-party authority. Compared with conventional many-to-one matching problems, the spectrum matching problem is distinctively challenging due to the interference bound between buyers: the same channel can be reused by an unlimited number of non-interfering buyers, but must be exclusively occupied by only one of interfering buyers. In this paper, we firstly formulate the spectrum matching problem as a many-to-one matching with peer effects, i.e., a buyer's utility is affected by other buyers who are matched to the same seller. We then present a two-stage distributed algorithm that converges to an interference-free and Nash-stable matching result. Simulations show that the proposed distributed matching algorithm can achieve 90% of the social welfare from the optimal matching result.

In a crowdsourcing system, it is important for the crowdsourcer to engineer extrinsic rewards to incentivize the participants. With mobile social networking, a user enjoys an intrinsic benefit when she aligns her behavior with the behavior of others. Referred to as network effects, such an intrinsic benefit becomes more significant as the number of users grows in the crowdsourcing system. But should a crowdsourcer design her extrinsic rewards differently when such network effects are taken into account? In this paper, we, for the first time, consider network effects as a contributing factor to intrinsic rewards, and study its influence on the design of extrinsic rewards. Rather than assuming a fixed participant population, we show that the number of participating users evolves to a steady equilibrium, thanks to subtle interactions between intrinsic rewards due to network effects and extrinsic rewards offered by the crowdsourcer. Taken network effects into consideration, we design progressively more sophisticated extrinsic reward mechanisms, and propose new and optimal strategies for a crowdsourcer to obtain a higher utility. Via extensive simulations, we demonstrate that with our new strategies, a crowdsourcer is able to attract more participants with higher contributed efforts; and participants gain higher utilities from both intrinsic and extrinsic rewards.

Dynamic video streaming protocols allow video clients to adaptively choose video rates to improve viewer Quality-of- Experience (QoE). Existing works on rate adaptation choose the video rate according to network capacity or buffer state, however, they didn't consider the early quitting problem. Most viewers quit before the video ends, in which case optimal solutions for complete video sessions may become suboptimal. In this paper, we design a video rate adaptation scheme which considers incomplete video sessions and aims at maximizing the quality of the video chunks that are actually watched by viewers. We formulate the video rate adaptation problem as a Markov Decision Process (MDP), in which the viewers' decision of quitting depends on the video quality and the rebuffer time. To solve the MDP problem, we propose two online learning algorithms for regular viewers and new viewers with no viewing history, respectively. The simulation results show that the proposed scheme increases the video quality {truly} consumed by the viewers, thus improving the viewer QoE.

Reflective wrist photoplethysmograph (PPG), obtained by a watch or wristband, can provide a natural and unconstrained way for daily life heart rate monitoring. However, reflective wrist PPG often suffers from poor signal quality and various distortions due to daily life motion artifact. In this paper, we analyze the influence of motion artifact on reflective wrist PPG signals, and propose a method to extract reliable heart rate from such distorted PPG signals. The proposed method consists of adaptive filtering, heart rate selection, and motion identification. Experimental results show that our proposed method can generate reliable heart rate values from wrist PPG signals with different types of motion artifact.

Major cellular operators are planning to upgrade to high-speed 4G networks, but due to budget constraints, they have to dynamically plan and deploy the 4G networks through multiple stages of time. By considering one-time deployment cost, daily operational cost and 3G network congestion, this paper studies how an operator financially manages the cash flow and plans the 4G deployment in a finite time horizon to maximize his final-stage profit. The operator provides both the traditional 3G service and the new 4G service, and we show that users will start to use the 4G service only when it reaches a sizable coverage. At each time stage, the operator first decides an additional 4G deployment size, by predicting users' responses in choosing between the 3G and 4G services. We formulate this problem as a dynamic programming problem, and propose an optimal threshold-based 4G deployment policy. We show that the operator will not deploy to a full 4G coverage in an area with low user density or high deployment/operational cost. Perhaps surprisingly, during the 4G deployment process, we show that the 4G subscriber number first increases and then decreases, as the 4G service helps mitigate 3G network congestion and increases its QoS.

To match wireless users' soaring traffic demand, spectrum regulators are considering allocating additional spectrum to the wireless market. There are two major directions for the spectrum allocation: licensed (e.g., 4G cellular service) and unlicensed services (e.g., Super Wi-Fi service). The 4G service provides a ubiquitous coverage, has a higher spectrum efficiency, and often charges users a high service price. The Super Wi-Fi service has a limited coverage, a lower spectrum efficiency, but often charges users a low service price. The spectrum regulator now simply allocates the spectrum to maximize its income, but such an income-centric allocation does not ensure the best spectrum utilization by the users. This motivates us to design a new spectrum allocation scheme which jointly considers the spectrum regulator's income and the users' aggregate utility by investigating three market tiers: the spectrum regulator, 4G and Super Wi-Fi operator coalitions, and all the wireless users. We formulate it as a three-stage game and derive the unique subgame perfect equilibrium. Compared with the traditional income-centric allocation, we prove that the proposed scheme significantly improves users' aggregate utility with a limited spectrum regulator's income loss.

In the spectrum auction, if a buyer locates in a “critical” place, interfering with a lot of other buyers, his occupancy of the spectrum may deprive many other transmission opportunities. In this paper, we propose a Location-aware Online Truthful doUble auction Scheme (LOTUS), which incorporates the buyers' location information into auction mechanism design. In the online auction, the biggest challenge is how to allocate the spectrums based on the knowledge in the current time slot, without knowing the spectrum requests that may come afterward. To solve this problem, we propose considering the opportunity cost of allocating the spectrum to a buyer based on his local interference conditions. We introduce the “interference discount” to markdown a buyer's bid if he induces a wide range of interference. Furthermore, we take into account the spectrum heterogeneity and design mechanisms that guarantee the economic robustness of the auction. The simulation results show that LOTUS outperforms the existing online auction mechanism, significantly improving buyers' and sellers' utility.

For wireless service providers (WSP), the emerging femtocell market brings new opportunities as well as challenges. It is difficult for a WSP to have complete information of the technical strength of other WSP in the market, and it is hard to estimate the user demand, which is influenced by the price and service quality of all competing WSP in the market. To address these problems, in this paper, we propose an economic framework for the WSP to maximize their utility, via a joint pricing and spectrum allocation strategy, under the condition of incomplete information of other rival WSP in the market. We study two scenarios: 1) all WSP enter the market at the same time; 2) the WSP enter the market at different times. In both scenario, we formulate the problem as a Bayesian game, and derive the Bayesian Nash equilibrium. The simulation results verify that the proposed joint pricing and spectrum allocation strategy outperforms sole pricing or sole spectrum allocation strategy. Interestingly, when the WSP enter the market at different times, the later entered WSP chooses more aggressive pricing and spectrum allocation strategy than the early entered WSP.

To accommodate the soaring mobile broadband traffic, the Federal Communications Commission (FCC) in the U.S. sets out to retrieve under-utilized spectrum (e.g., TV Whitespace) and lay the groundwork for spectrum redistribution. Auction is an efficient way to allocate resources to those who value them the most. The large pool of spectrums to be released, especially the ones in TV Whitespace, consist of wide-range frequencies. Apart from spatial reuse, spectrum heterogeneity imposes new challenges for spectrum auction design: 1) Wireless service providers with different targeted cell coverages have different spectrum frequency preferences; 2) interference relationship is frequency-dependent due to frequency-selective signal fading. Unfortunately, existing spectrum auction mechanisms either assume spectrum valuation is homogeneous or use homogeneous interference graph to group buyers who can reuse the same spectrum. In this paper, we propose TAMES, an auction framework for heterogeneous spectrum transaction. We consider a multi-seller-multi-buyer double auction, in which every buyer submits a bid, consisting of the spectrum demand and a bidding profile of prices for spectrums contributed by all sellers. A novel buyer grouping approach is proposed to tackle the problem of heterogeneous interference graph. TAMES is proved to be truthful as well as individually rational. The simulation results show that TAMES significantly improves spectrum utilization, sellers' revenue and buyers' utility by making smart use of spectrum heterogeneity, while keeping low running time comparable with existing auction mechanisms. Moreover, via simulation, we show how to help buyers obtain continuous spectrums which further improves buyers' satisfaction.

With the prosperity of the Internet, many advertisers choose to deliver their advertisements by online targeting, where the ad broker is responsible for matching advertisements with users who are likely to be interested in the underlying products or services. However, this online advertisement targeting system requires user profile information and may fail due to privacy issues. In light of growing privacy concerns, we propose a privacy-aware framework for online advertisement targeting, where users are compensated for their privacy leakage and motivated to click more advertisements. In the framework, an ad broker pays a varying amount of money to users for clicking different advertisements due to distinct privacy leakage. Meanwhile advertisers send advertisements to the ad broker and determine the price per user click they need to pay. We model the interactions among advertisers, the ad broker and users as a three-stage game, where every player aims at maximizing its own utility, and Nash Equilibrium is achieved by backward induction. We further analyze the optimal strategies for advertisers, the ad broker and users. Numerical results have shown that the proposed privacy-aware framework is effective as it enables all advertisers, the ad broker and users to maximize their utilities in case of different levels of user privacy sensitivities. In addition, the proposed framework produces higher profits for advertisers and the ad broker than the traditional “paid to click” system.

Femtocell refers to a new class of low-power, low-cost base stations (BSs) which can provide improved indoor coverage and higher voice/data Quality of Service (QoS). Hybrid access in two-tier macro-femto networks is regarded as the most ideal access control mechanism to help offload macrocell traffic to femtocell, thus enhancing overall network performance. However, without suitable incentive mechanism, the Femtocell Service Providers (FSPs) are not willing to share their femtocell resource with the Macrocell Service Provider (MSP). To address this problem, in this paper, we propose an ACcess Permission (ACP) transaction framework, in which a single MSP purchases ACP from multiple FSPs in various locations throughout T timeslots, and FSPs who have overlapped coverage compete with each other for selling their ACP. However, we are facing the challenge that the demand of MSP in each location dynamically changes at each timeslot. At the start of each timeslot, FSPs are unaware of the demand of MSP, which impedes them to choose an ideal strategy that yields high payoff. To address the problem of information incompleteness, we propose an adaptive strategy updating algorithm, which is based on online learning process and enables FSPs to obtain guaranteed payoff. We conduct simulations to evaluate the payoff and the payoff gap of the FSPs when the MSP's demand is constant, quasi-constant or probabilistic. We also show that the payoff of the FSPs is affected by the learning speed of the proposed algorithm.

To accommodate users' ever-increasing traffic in wireless broadband services, the Federal Communications Commission (FCC) in the U.S. is considering allocating additional spectrum to the wireless market. There are two major directions: licensed (e.g. 3G) and unlicensed services (e.g. Wi-Fi). On the one hand, 3G service can realize a high spectrum efficiency and provide ubiquitous connection. On the other hand, the Wi-Fi service (often with limited coverage) can provide users with high-speed local connections, but is subject to uncontrollable interferences. Regarding spectrum allocation, prior studies only focused on revenue maximization. However, one of FCC's missions is to better improve all wireless users' utilities. This motivates us to design a spectrum allocation scheme that jointly considers social welfare and revenue. In this paper, we formulate the interactions among the FCC, typical 3G and Wi-Fi operators, and the endusers as a three-stage dynamic game and derive the equilibrium of the entire game. Compared to the benchmark case where the FCC only maximizes its revenue, the consideration of social welfare will encourage the FCC to allocate more spectrum to the service which lacks spectrum to better serve its users. Such consideration for the social welfare, to our delight, brings limited revenue loss for the FCC.

Spectrums are heterogeneous, especially from the aspect of their central frequency. According to signal propagation properties, low-frequency spectrum generally has lower path loss, thus longer transmission range, compared with high-frequency spectrum. Cellular operators with different targeted cell size will have different preferences for spectrums with different frequencies. Furthermore, the transmission range also affects the interference relationships among transmitters. Transmitters who can reuse the same high-frequency spectrum may interfere with each other when reusing the low-frequency spectrum, so it is difficult to decide how to construct the interference graph to exploit spectrum reusability among transmitters. Auction is considered as an efficient way for spectrum allocation. However, most of the previous works only considered homogenous spectrum auction, failing to address the problem of spectrum heterogeneity. In this paper, we propose TAMES, a Truthful Auction Mechanism for hEterogeneous Spectrum allocation, which allows buyers to freely express their different preferences towards different spectrums. Frequency-specific interference graphs are constructed to determine buyer groups. The proposed heterogeneous spectrum auction is theoretically proved to be truthful and individual rational. The simulation results verifies that the proposed auction mechanism outperforms other auction mechanisms with homogenous bid or homogenous interference graph. The proposed auction mechanism is able to yield higher buyers' satisfaction, seller's revenue and spectrum utilization.

Information platform is managed by IPSPs (Information Platform Service Providers) who aggregate and compile information from a large pool of source websites supported by Source Service Providers (SSPs). While information platform makes it much more convenient for users to browse information, there exists such an economic dilemma: Information platform may send extra user clicks to source websites (thus increasing SSPs' advertising revenue), some users get stuck on the information platform since information excerpts are presented. In this case, SSPs tend to accuse IPSPs of intellectual property infringement. To boost the development of information platform, it is better to make IPSPs and SSPs allies rather than enemies. In this paper, we propose a business framework which involves a single IPSP and multiple SSPs. Within the framework, IPSP gets paid for the user clicks sent from information platform to the source websites and SSPs donate certain amount of information to information platform. We formulate the framework as a twostage game, assuming that IPSP and SSPs are selfish and rational players who target at utility maximization. We use backward induction to get Nash Equilibrium of the game and give the best strategies for IPSP and SSPs. The numerical results have shown that the enhancement of information quality provided by source websites and increased non-material profit of information platform will improve IPSP's utility. In addition, the price charged by IPSP for user clicks and aggregated information contributed by SSPs are also affected by these factors.

Femtocell refers to a new class of low-power, low-cost base stations (BSs) which can provide better coverage and improved voice/data Quality of Service (QoS). Hybrid access in two-tier macro-femto network is regarded as the most ideal access control mechanism to enhance overall network performance. But the implementation of hybrid access is hindered by a lack of market that can motivate ACcess Permission (ACP) trading between Wireless Service Providers (WSPs) and private femtocell owners. In this paper, we propose a reverse auction framework for fair and efficient ACP transaction. Unlike strict outcome (the demand of bidder must be fully satisfied) in most of the existing works on auction design, the proposed auction model allows range outcome, in which WSP accepts partial demand fulfillment and femtocell owners makes best-effort selling. We first propose a Vickery-Clarke-Grove (VCG) based mechanism to maximize social welfare. As the VCG mechanism is too time-consuming, we further propose an alternative truthful mechanism (referred to as suboptimal mechanism) with acceptable polynomial computational complexity. The simulation results have shown that the suboptimal mechanism generates almost the same social welfare and the cost for WSP as VCG mechanism.

Auction is widely applied in wireless communication for spectrum allocation. Most of prior works have assumed that spectrums are identical. In reality, however, spectrums provided by different owners have distinctive characteristics in both spacial and frequency domains. Spectrum availability also varies in different geo-locations. Furthermore, frequency diversity may cause non-identical conflicts among spectrum buyers since different frequencies have distinct communication ranges. Under such realistic scenario, existing spectrum auction schemes cannot provide truthfulness or efficiency. In this paper, we propose a Truthful double Auction for HEterogeneous Spectrum, called TAHES. TAHES allows buyers to explicitly express their personalized preferences for heterogeneous spectrums and also addresses the problem of interference graph variation. We prove that TAHES has nice economic properties including truthfulness, individual rationality and budget balance.

Femtocell technology addresses the problem of poor indoor coverage, benefiting both wireless service provider (WSP) and end users. With the introduction of femtocell, the cross-tier interference between macro link and femto link becomes a major factor which greatly impacts the network performance. Different access control approaches, by generating different interference patterns, also severely affect the overall throughput of the network and need to be carefully investigated. Among all the access control mechanisms, hybrid access is the most promising one, which allows roaming unregistered users (referred to as macro users) to access the nearby femto base station (BS) while reserving certain resource for registered home users (referred to as femto users), improving overall network capacity. However, to successfully leverage hybrid access is challenging because the femto holders (FHs) are selfish, unwilling to share their femto facilities and spectrum resource with macro users without any incentive mechanism. In this paper, we propose a novel utility-aware refunding framework to motivate hybrid access in femtocell. Within the framework, both WSP and FHs are assumed to be selfish, and target at maximizing their own utilities. WSP provides certain refunding to motivate FHs to open their resource for macro users. FHs decide the resource allocation among femto and macro users according to the amount of refunding WSP offers. Under this framework, the optimal strategies of both WSP and FHs are analyzed by formulating the problem as a Stackelberg Game. A unique Nash Equilibrium is achieved and a hybrid access protocol is designed according to the analysis. Extensive simulations have been conducted and the results show that the utilities of both WSP and FHs are significantly improved exploiting the hybrid access mechanism.

Auction is widely applied in wireless communication for spectrum allocation. Most of prior works have assumed that all spectrums are identical. In reality, however, spectrums provided by different owners have distinctive characteristics in both spacial and frequency domains. Spectrum availability also varies in different geo-locations. Furthermore, frequency diversity may cause non-identical conflict relationships among spectrum buyers since different frequencies have distinct communication ranges. Under such a scenario, existing spectrum auction schemes cannot provide truthfulness or efficiency. In this paper, we propose a Truthful double Auction mechanism for HEterogeneous Spectrum, called TAHES, which allows buyers to explicitly express their personalized preferences for heterogeneous spectrums and also addresses the problem of interference graph variation. We prove that TAHES has nice economic properties including truthfulness, individual rationality and budget balance. Results from extensive simulation studies demonstrate the truthfulness, effectiveness and efficiency of TAHES.

Femtocell technology is regarded as a promising way to deal with poor indoor coverage and increase spectrum spatial reuse. In this paper, we focus on the scenario that macro and femto base stations are deployed by the same Wireless Service Provider (WSP), which treats the revenue maximization as its ultimate target. In such a system, there are several design factors which will affect the overall revenue, including price decision and resource allocation between macrocell and femtocell. In this paper, we propose an economic framework, where users choose either macrocell or femtocell service to optimize their own utility and the monopolistic WSP tries to maximize its revenue via pricing and spectrum allocation strategy. Theoretical results of optimal prices for macrocell and femtocell are given. Extensive theoretical analysis is carried out to determine the spectrum allocation strategy and evaluate the revenue of the WSP. The system capacity and the ratio of macrocell and femtocell users are also discussed. The results have indicated that the revenue of the WSP is significantly improved by combining the pricing strategy and the spectrum allocation strategy.