The vector database stores data as high-dimensional feature vectors. Some recently proposed attack techniques enable an adversary to launch feature vector inversion (FVI) attacks against vector databases. In FVI attacks, an adversary trains an FVI attack network to reconstruct the original private data from their feature vectors based on the assumption that an auxiliary dataset is available to the adversary. However, such a data-available assumption is too strong, making such FVI attacks unrealistic in many real-world scenarios. In this paper, we make the first systematic study on FVI attacks against vector databases in the data-free setting. To tackle the issue of no training data, we develop an output-to-input data generation technique that helps to generate synthetic fake samples for the FVI attack network training. In addition, to ensure the high quality of generated fake samples, we develop the accelerable complete bipartite graph (CBG) search strategy and the downstream-classifier-aided generator training strategy. Furthermore, as the key insight of this work, we find that the proposed output-to-input data generation technique can be employed to launch the other three ML attacks. Intriguingly, we find that the proposed FVI attack technique in the data-free setting can be directly employed to boost the attack performance of FVI attacks in the auxiliary-dataset-available setting. Finally, we propose and study defenses against the proposed attacks.

Transfer learning (TL) by fine-tuning (FT) has become a popular paradigm to address the challenges of limited training data and computing resources encountered in model training. This study reveals that this paradigm is susceptible to a new threat called model parasitic (MP) attack. By poisoning the dataset used for fine-tuning, MP attacks enable the finetuned model to execute an additional task (e.g., a specific classification task) designated by the attacker while still being able to execute the original task that the victim's fine-tuned model aims to offer. In addition, through MP attacks, the attacker can free-ride the victim's machine learning (ML) services at the cost of the victim. To design MP attacks, we innovatively propose multiple strategies including the dual-cluster strategy, the benign-to-poisoned example generation strategy, and the feature assignment loss (FAL)-guided controllable perturbation search strategy. Through intensive experiments, we precisely identify the factors that influence the MP attack performance. Finally, we investigate three possible defenses, shedding light on more effective defense design. Our code is available at GitHub

Smart homes, medical devices, and education systems, among other emerging cyber-physical systems, hold immense promise for sensing-based user interfaces, especially for using fingers and hand gestures as system input. However, vision approaches compatible with time-consuming image processing adopt low 60 Hz location sampling rate (frame rate) for real-time hand gesture recognition. Furthermore, they are not suitable for low-light environment and long detection range. In this paper, we propose RoFin, which first exploits 6 temporal-spatial 2D rolling fingertips for real-time 3D reconstructing of 20-joint hand pose. RoFin designs active optical labeling for finger identification and enhances inside-frame 3D location tracking via high rolling shutter rate (5-8 kHz). These features enable great potentials for enhanced multi-user HCI, virtual writing for Parkinson suffers, etc. We implement RoFin prototypes with wearable gloves attached with low-power single-colored LED nodes and commercial cameras. The experiment results show that (1) In flexible sensing distances up to 2.5 m, RoFin achieves an average labeling parsing accuracy of 85%, (2) In comparison to vision-based techniques, RoFin improves the tracking grain with 4× more sampled points each frame, (3) RoFin reconstructs a hand pose in real time with 16 mm mean deviation error compared with Leap Motion under flexible distance, and (4) we further investigated real-world applications of RoFin, such as air writing with smoothed trajectories and mobile-based letter/number recognition in our developedXameraapp.

Door Access Control (DAC) plays a pivotal role in balancing security and convenience in modern infrastructures. However, current vision-based DAC systems face limitations such as privacy risks, degraded performance, and high computational costs, while RFID and Bluetooth alternatives remain vulnerable to replay, cloning, and eavesdropping attacks. Recent advances in visible light sensing and backscatter communication have enabled promising opportunities for secure, low-power, and cost-effective access control systems. We propose ViKey, the first visible light backscatter-based DAC system that utilizes polarized birefringence to generate 3D position-dependent color patterns as keys, enabling robust and contactless authentication. The design methodology of ViKey highlights the potential of visible light backscatter technology for future smart DAC applications.

Self-driving platooning trucks are becoming increasingly common due to the economic gains for companies that utilize them. However, high throughput, long range communication between truck platoons can be very difficult if they are not in areas with pre-built cellular infrastructures. To combat this problem, we propose PlaCoB (Platoon Collaborative Beamforming), which enables multiple platooning trucks to collaboratively beamform to maintain communication over long distances. We conduct extensive simulations under various settings to evaluate PlaCoB’s performance compared to single-truck methods. By leveraging multi-truck platoons, collaborative beamforming, and frequency shifting, (1) PlaCoB can transmit 1.23x more data compared to single vehicle transmission methods, and (2) PlaCoB can transmit data 2.44x further than single vehicle transmission methods. These results demonstrate the effectiveness of our proposed PlaCoB’s approach.

Door Access Control (DAC) plays a pivotal role in balancing security and convenience in modern infrastructures. However, current vision-based DAC systems exhibit limitations including privacy concerns (e.g., facial data leakage), performance degradation under suboptimal lighting conditions, high computational overhead and system costs. While RFID and Bluetooth-based alternatives exist, they exhibit vulnerabilities to attacks including replay attacks, signal cloning, and eavesdropping. Recent advances in visible light sensing and backscatter communication have enabled promising opportunities for secure, low-power access control systems with sub-dollar hardware costs. In this paper, we propose ViKey, the first visible light backscatter-based DAC system that utilizes polarized birefringence to generate 3D position-dependent color patterns as keys, enabling robust and contactless authentication. We design and implement a ViKey prototype using commercial off-the-shelf (COTS) components, with a tag cost of less than $0.2. Real-world experiments show that our current ViKey prototype can achieve an average authentication accuracy of 90.5% at 0.5m with our best patterns. These results demonstrate the effectiveness of our low-cost visible light backscatter technology for future smart DAC applications.

With the proliferation of Internet-of-Things and the advancements in deep learning and Artificial Intelligence, there is an increased need for task-specific mobile devices that can efficiently utilize the wireless spectrum while operating in a wide variety of environments and channel conditions. This work proposes a novel end-to-end Visual-Inertial Odometry architecture that utilizes Semantic Communication-enhanced modulation constellation design for pose estimation of mobile devices. Our proposed model occupies less wireless spectrum bandwidth and does not require complex forward error-correction mechanisms. Yet, it is able to infer the pose information of a remote mobile device with comparable accuracy to a standard wireless Visual-Inertial Odometry system. We propose utilizing a Vision Transformer-based Image-IMU encoder for a network-constrained remote mobile device that transmits wireless encoded data to an Edge receiver. The receiver automatically detects the modulation scheme and decodes the received data for pose estimation. We propose a novel modulation constellation coding design scheme that can transmit data in multiple Modulation and Coding Schemes (MCS) within the same burst, without the need to embed an MCS symbol within the burst. Our proposed receiver can automatically detect the MCS as well. We illustrate how our proposed architecture can transmit encoded Image-IMU data over a wide range of distances and Signal-to-Noise Ratio (SNR) channel conditions and have a decoded pose accuracy comparable to a standard wireless Visual-Inertial Odometry system, with significantly less wireless bandwidth consumption and computational complexity.

Optical wireless communication (OWC) has inherent location aware and spatial reuse advantages over RF-based technologies due to the Line-of-Sight (LoS) propagation of optical signals. Hence, OWC presents a fresh opportunity for effective and secure IoT connectivity and data transmission. However, most OWC systems design transmitter as a point source without considering its spatial diversity in data delivery in 3D space, which is not suitable for real-world mobile IoT connections among devices and users. In this paper, we design and implement HoloCube, which provides 3D optical IoT connections via software defined optical camera communication and Pepper's ghost effect. At the heart of the HoloCube design is multiple virtual 3D hollowed-out cubes with adaptive Spatial-Color Shift Keying (SCSK) modulation. Specifically, the virtual cube seen from various directions has constant structure but embeds different data over time. The cube's positioning elements provide double reference for both 3D reconstruction (spatial) and robust color decoding (spectral). Our comprehensive experiments demonstrate that HoloCube achieves practical 3D omnidirectional IoT connections with 70 Kbps goodput at 4 m in real-world indoor setting.

In this work, we present RadCloud, a novel real-time framework for directly obtaining higher-resolution lidar-like 2D point clouds from low-resolution radar frames on resource-constrained platforms commonly used in unmanned aerial and ground vehicles (UAVs and UGVs, respectively); such point clouds can then be used for accurate environmental mapping, navigating unknown environments, and other robotics tasks. While high-resolution sensing using radar data has been previously reported, existing methods cannot be used on most UAVs, which have limited computational power and energy; thus, existing demonstrations focus on offline radar processing. RadCloud overcomes these challenges by using a radar configuration with 1/4th of the range resolution and employing a deep learning model with 2.25× fewer parameters. Additionally, RadCloud utilizes a novel chirp-based approach that makes obtained point clouds resilient to rapid movements (e.g., aggressive turns or spins) that commonly occur during UAV flights. In real-world experiments, we demonstrate the accuracy and applicability of RadCloud on commercially available UAVs and UGVs, with off-the-shelf radar platforms on-board.

In this paper, we propose WavePurifier, an audio purification framework to defend against audio adversarial attacks. Audio adversarial attacks craft adversarial examples or perturbations to attack the automated speech recognition (ASR) models. Although existing defense mechanisms can detect such attacks and raise alarms, they fail to recover or maintain benign commands. Consequently, this leads to the denial of users' benign commands. Different than existing defenses, WavePurifier aims to purify adversarial examples, thereby rectifying the user's benign commands. We find that the forward diffusion process of the diffusion model effectively eliminates perturbations, whereas the reverse diffusion process restores benign speech. Based on this, we develop a hierarchical diffusion model to defend against audio adversarial examples. This model is capable of purifying different spectrogram bands to varying degrees. To validate the performance of WavePurifier, we purify the adversarial examples from 3 different adversarial attacks in 140 distinct settings. In total, we collect 78,864 diffused spectrograms and 21,000 purified audios. Then, we evaluate WavePurifier on 2 different ASR models, 4 commercial speech-to-text APIs, 2 real-world attack scenarios, and compare them against 7 existing defense approaches. Our result shows that WavePurifier is a universal framework, demonstrating adaptability across diverse attacks with the same hyperparameters. Notably, WavePurifier outperforms existing methods with the lowest character error rate (CER), word error rate (WER), and a high purification success rate against different attacks.

We demonstrate RadCloud, a real-time framework for obtaining high-resolution lidar-like 2D point clouds from low-resolution millimeter-wave (mmWave) radar data on resource-constrained platforms commonly found on unmanned aerial and ground vehicles (UAVs and UGVs). Such point clouds can then be used for mapping key features of the environment, route planning and navigation, and other robotics tasks. Rad-Cloud is specifically optimized for UAVs and UGVs by using a radar configuration with 1/4th the range resolution, using a model with 2.25× fewer parameters, and reducing total sensing time by a factor of 250×. The real-time ROS framework will be demonstrated on a UGV and UAV equipped with CPU-only compute platforms in diverse environments.

Federated learning (FL) supports distributed training of a global machine learning model across multiple Internet of Things (IoT) devices with the help of a central server. However, data heterogeneity across different IoT devices leads to the client model drift issue and results in model performance degradation and poor model fairness. To address the issue, we design federated learning with global–local knowledge fusion (FedKF) scheme in this article. The key idea in FedKF is to let the server return the global knowledge to be fused with the local knowledge in each training round so that the local model can be regularized toward the global optima. Therefore, the client model drift issue can be mitigated. In FedKF, we first propose the active–inactive model aggregation technique that supports a precise global knowledge representation. Then, we propose a data-free knowledge distillation (KD) approach to enable each client model to learn the global knowledge (embedded in the global model) while each client model can still learn the local knowledge (embedded in the local data set) simultaneously, thereby realizing the global–local knowledge fusion process. The theoretical analysis and intensive experiments demonstrate the superiority of FedKF over previous solutions.

Optical wireless communication (OWC) shows great potential due to its broad spectrum and the exceptional intensity switching speed of LEDs. Under poor conditions, most OWC systems switch from complex and more error prone high-order modulation schemes to more robust On-Off Keying (OOK) modulation defined in the IEEE OWC standard. This paper presents LiFOD, a high-speed indoor OOK-based OWC system with fine-grained dimming support. While ensuring fine-grained dimming, LiFOD remarkably achieves robust communication at up to 400 Kbps at a distance of 6 meters. This is the first time that the data rate has improved via OWC dimming in comparison to the previous approaches that consider trading off dimming and communication. LiFOD makes two key technical contributions. First, LiFOD utilizes Compensation Symbols (CS) as a reliable side-channel to represent bit patterns dynamically and improve throughput. We firstly design greedy-based bit pattern mining. Then we propose 2D feature enhancement via YOLO model for real-time bit pattern mining. Second, LiFOD synchronously redesigns optical symbols and CS relocation schemes for fine-grained dimming and robust decoding. Experiments on low-cost Beaglebone prototypes with commercial LED lamps and the photodiode (PD) demonstrate that LiFOD significantly outperforms the state-of-the-art system with 2.1× throughput on the SIGCOMM17 data-trace.

Optical Camera Communication (OCC) appears as a promising technology to provide secure and pervasive wireless services with users' daily smart devices. Rolling shutter based modulations can improve the frequency response of the camera. This paper introduces a 2D Rolling Block (2DRB) based OCC modulation to use un-exploited spatial diversity to improve OCC's data rate for real-world applications. 2DRB outperforms traditional 1D strip based modulations. Using our 2DRB prototype with commercial devices, we show a significant data rate enhancement. We also discuss one promising real-world use case: indoor office integrated lighting and communication.

Smart homes, medical devices, and education systems, among other emerging cyber-physical systems, hold immense promise for sensing-based user interfaces, especially for using fingers and hand gestures as system input. However, vision approaches compatible with time-consuming image processing adopt low 60 Hz location sampling rate (frame rate) for real-time hand gesture recognition. Furthermore, they are not suitable for low-light environment and long detection range. In this paper, we propose RoFin, which first exploits 6 temporal-spatial 2D rolling fingertips for real-time 3D reconstructing of 20-joint hand pose. RoFin designs active optical labeling for finger identification and enhances inside-frame 3D location tracking via high rolling shutter rate (5--8 KHz). These features enable great potentials for enhanced multi-user HCI, virtual writing for Parkinson suffers, etc. We implement RoFin prototypes with wearable gloves attached with low-power single-colored LED nodes and commercial cameras. The experiment results show that (1) In flexible sensing distances up to 2.5 m, RoFin achieves an average labeling parsing accuracy of 85%. (2) In comparison to vision-based techniques, RoFin improves the tracking grain with 4× more sampled points each frame. (3) RoFin reconstructs a hand pose in real time with 16 mm mean deviation error compared with Leap Motion under flexible distance.

Drones are gaining more interest thanks to their advantages and great potential for applications. However, present swarming drones’ stand-alone centralized radio frequency control mode from a base station has non-trivial drawbacks such as severe interference, latency caused localization error, etc. Differently, optical camera communication (OCC) is promising as an alternative for integrated communication and sensing for swarming drones. We propose PoseFly, the first 4-in-1 OCC approach for swarming drones. With exploited rolling shutter effect and the already installed camera and LED nodes, PoseFly provides (1) massive drone indication and identification, (2) multilevel on-site localization, (3) quick-link channel among drones, and (4) basic lighting. The design methodology of PoseFly gives a valuable example for the low-cost integrated sensing and communication for swarming drones.

Drones are gaining more interest from the industry and the research community as a result of their many advantages, including low cost, small size, adaptability, and ease of use, as well as their potential applications. However, current control of swarming drones relies on stand-alone modes and centralized radio frequency control from a base station on the ground which is devoid of drone-to-drone communication. This method has drawbacks, including a crowded RF spectrum with mutual interference, high latency, and a lack of on-site drone-to-drone interactions. Because of its high spatial multiplexing capability, Line of Sight (LoS) security capabilities, broader bandwidth, and intuitive vision manner, Optical Camera Communication (OCC) is considered to be a potential alternative for sensing and communication in drone clusters. In this paper, we first utilize the rolling shutter effect in drone sensing and communication and propose PoseFly, a 4-in-1 AI-assisted OCC with drone identification, on-site localization, quick-link communication and lighting. We implement PoseFly prototypes on commercial drones, cameras and LEDs. Our experiments show our PoseFly achieves nearly 100% accuracy for distance estimation (20m), drone identification (12m), angle and speed estimation (4m) and 5 Kbps average quick-link throughput at up to 4 m on current prototypes.

Underwater optical wireless communication techniques are promising due to a broad bandwidth with a long communication range compared with existing expensive acoustic and RF-based underwater communication techniques. For underwater navigation assistance during dive and rescue, it is more practical to adopt passive optical tags for objects/human identification and location-based services. However, existing optical tags (bar/QR codes) employ one/two dimensional designs, which lack significant element/symbol distance for robust decoding and full-directional localization capabilities for underwater navigation tasks. This paper investigates opportunities to increase the element distance in passive low-order optical tags by exploiting 3D spatial diversity. Specifically, we design U-Star, a system that consists of Underwater Optical Identification (UOID) tags and commercial camera-based tag readers for underwater navigation. Our UOID tags embed rich location and guidance information. Additionally, because our UOID tags employ a three-dimensional design, they can also determine the relative location of a user in real-time based on the perspective principles. We design AI based mobile algorithms for underwater denoising, relative positioning, and robust data parsing for tag readers. Finally, we evaluate U-Star on real UOID tag prototypes under different underwater scenarios. Results show that our 3-order UOID tag can embed 21 bits with a BER of 0.003 at 1m and less than 0.05 at up to 3m, which is sufficient for underwater navigation guidance with backup database.

Optical wireless communication (OWC) shows great potential for high-speed communication due to its broad spectrum and the exceptional intensity switching speed of LEDs. Under poor conditions, most OWC systems switch from complex and more error prone high-order modulation schemes to the more robust On-Off Keying (OOK) modulation defined in the IEEE OWC standard. This paper presents LiFOD, a high-speed indoor OOK-based OWC system with fine-grained dimming support. While ensuring fine-grained dimming, LiFOD remark-ably achieves robust communication at up to 400 Kbps at a distance of 6 meters. This is the first time that the data rate has improved via OWC dimming in comparison to the previous approaches that consider trading off dimming and communication. LiFOD makes two key technical contributions. First, LiFOD utilizes Compensation Symbols (CS) as a reliable side-channel to represent bit patterns dynamically and improve throughput. Second, LiFOD synchronously redesigns optical symbols and CS relocation schemes for fine-grained dimming and robust decoding. Experiments on low-cost Beaglebone prototypes with commercial LED lamps and the photodiode (PD) demonstrate that LiFOD significantly outperforms the state-of-art system with at least 2.1x throughput on the SIGCOMM17 data-trace.

Sensing-based user interfaces hold enormous potential for smart homes, medical equipment, educational systems, AR/VR/MR, etc. However existing hand and body gesture recognition systems are mostly based on frame-level computer vision approaches, which have limitations such as the inability to operate in the environment with low brightness, short detection distance, without the objective identification ability, and coarse-grained tracking when the objectives are in high-speed motion. Therefore, in this paper, we propose to attach active LED elements on objectives and utilize rolling shutter effect to enhance the gesture recognition and achieve the fine-grained motion tracking with objective identification.

Thanks to the popularity of smartphones, optical camera communication (OCC) gains more and more attention from markets and research as one of the new internetworking architectures in the future. However, existing OCC modulations treat the signal from the transmitter as a single point light source, which sacrifices the spatial diversity and limits the data rate improvement. In this paper, we investigate the spatial diversity, a natural feature of camera imaging, and propose to combine spatial diversity with amplitude and spectrum diversities to boost the data rate of OCC to support high-speed applications. Based on preliminary experiments, we design and implement RainbowRow, a high-speed and robust OCC system. RainbowRow comprises a LED-based transmitter and a camera-based receiver, which is low-cost (i.e., under $100), with a flexible communication range (i.e., within 2 m), and high speed (i.e., up to 40 Kbps). To validate the RainbowRow performance, we conduct experiments on LabVIEW based platforms in different scenarios. Results show that RainbowRow can robustly recognize the optical symbols with a symbol error rate (SER) less than 0.05 and approach 40 Kbps data rate in the range of 1 m, significant improvement from less than 10 Kbps in OCC state of the art.

The emerging 5G adopts OFDM modulation and deploys small-cell amplify and forward (AF) relays for in-cell capacity enhancement and energy efficiency. However, hybrid delivery services for multiple users where broadcast and unicast coexist are inefficient and unfair due to their different QoS requirements. Most existing work considering hybrid broadcast and unicast traffic focuses on different scheduling schemes in onehop scenarios. For dual-hop relay networks, subcarrier mapping or pairing has been studied, but none considers hybrid traffic with both broadcast and unicast. In this paper, we propose an effective subcarrier pairing (ESP) protocol, which exploits the performance diversity in subcarrier pairing at relays to improve the overall performance of hybrid broadcast and unicast traffic. ESP exquisitely pairs subcarriers of two hops into two kinds of subcarrier pairs separately. ESP then allocates subcarrier pairs with low outage probability for broadcast and subcarrier pairs with high capacity for unicast. In ESP design, we study several important metrics such as end-to-end outage probability, capacity, and bit-error-rate (BER) in AF assisted OFDM-CDMA relay networks. We conduct Monte Carlo simulations to verify the effectiveness and fairness of our approach in hybrid transmission. Results show that ESP is efficient in hybrid delivery for relay networks and improves the performance of broadcast services significantly without sacrificing unicast services.

Unmanned Aerial Vehicles (UAVs) have wide applications in many fields, e.g. multiple Unmanned Aerial Vehicles (UAVs) cooperatively tracking multiple targets. This paper studies the multiple UAVs cooperatively tracking multiple targets by vision surveillance system, where the images/videos of targets have direction requirements. One target is covered by a UAV if and only if its position is within the Field Of View (UAV) as well as the UAV is within a requested angle of the target's face direction. The objective is to maximize the total covered targets number by the UAVs, which can not be solved by existing models. A simple effective distributed, online cooperation algorithm for this problem is designed in this paper. The theoretical analysis shows our algorithm achieves constant factor to the optimal.