3D Human Pose Estimation (3D HPE) is vital in various applications, from person re-identification and action recognition to virtual reality. However, the reliance on annotated 3D data collected in controlled environments poses challenges for generalization to diverse in-the-wild scenarios. Existing domain adaptation (DA) paradigms like general DA and source-free DA for 3D HPE overlook the issues of non-stationary target pose datasets. To address these challenges, we propose a novel task named lifelong domain adaptive 3D HPE. To our knowledge, we are the first to introduce the lifelong domain adaptation to the 3D HPE task. In this lifelong DA setting, the pose estimator is pretrained on the source domain and subsequently adapted to distinct target domains. Moreover, during adaptation to the current target domain, the pose estimator cannot access the source and all the previous target domains. The lifelong DA for 3D HPE involves overcoming challenges in adapting to current domain poses and preserving knowledge from previous domains, particularly combating catastrophic forgetting. We present an innovative Generative Adversarial Network (GAN) framework, which incorporates 3D pose generators, a 2D pose discriminator, and a 3D pose estimator. This framework effectively mitigates domain shifts and aligns original and augmented poses. Moreover, we construct a novel 3D pose generator paradigm, integrating pose-aware, temporal-aware, and domain-aware knowledge to enhance the current domain's adaptation and alleviate catastrophic forgetting on previous domains. Our method demonstrates superior performance through extensive experiments on diverse domain adaptive 3D HPE datasets.

Recent advances in dance generation have enabled the automatic synthesis of 3D dance motions. However, existing methods still face significant challenges in simultaneously achieving high realism, precise dance-music synchronization, diverse motion expression, and physical plausibility. To address these limitations, we propose a novel approach that leverages a generative masked text-to-motion model as a distribution prior to learn a probabilistic mapping from diverse guidance signals, including music, genre, and pose, into high-quality dance motion sequences. Our framework also supports semantic motion editing, such as motion inpainting and body part modification. Specifically, we introduce a multi-tower masked motion model that integrates a text-conditioned masked motion backbone with two parallel, modality-specific branches: a music-guidance tower and a pose-guidance tower. The model is trained using synchronized and progressive masked training, which allows effective infusion of the pretrained text-to-motion prior into the dance synthesis process while enabling each guidance branch to optimize independently through its own loss function, mitigating gradient interference. During inference, we introduce classifier-free logits guidance and pose-guided token optimization to strengthen the influence of music, genre, and pose signals. Extensive experiments demonstrate that our method sets a new state of the art in dance generation, significantly advancing the quality and editability over existing approaches.

Human mesh recovery (HMR) is crucial in many computer vision applications; from health to arts and entertainment. HMR from monocular images has predominantly been addressed by deterministic methods that output a single prediction for a given 2D image. However, HMR from a single image is an ill-posed problem due to depth ambiguity and occlusions. Probabilistic methods have attempted to address this by generating and fusing multiple plausible 3D reconstructions, but their performance has often lagged behind deterministic approaches. In this paper, we introduce GenHMR, a novel generative framework that reformulates monocular HMR as an image-conditioned generative task, explicitly modeling and mitigating uncertainties in the 2D-to-3D mapping process. GenHMR comprises two key components: (1) a pose tokenizer to convert 3D human poses into a sequence of discrete tokens in a latent space, and (2) an image-conditional masked transformer to learn the probabilistic distributions of the pose tokens, conditioned on the input image prompt along with randomly masked token sequence. During inference, the model samples from the learned conditional distribution to iteratively decode high-confidence pose tokens, thereby reducing 3D reconstruction uncertainties. To further refine the reconstruction, a 2D pose-guided refinement technique is proposed to directly fine-tune the decoded pose tokens in the latent space, which forces the projected 3D body mesh to align with the 2D pose clues. Experiments on benchmark datasets demonstrate that GenHMR significantly outperforms state-of-the-art methods.

The introduction of vision-language models like CLIP has enabled the development of foundational video models capable of generalizing to unseen videos and human actions. However, these models are typically trained on web videos, which often fail to capture the challenges present in Activities of Daily Living (ADL) videos. Existing works address ADL-specific challenges, such as similar appearances, subtle motion patterns, and multiple viewpoints, by combining 3D skeletons and RGB videos. However, these approaches are not integrated with language, limiting their ability to generalize to unseen action classes. In this paper, we introduce SKI models, which integrate 3D skeletons into the vision-language embedding space. SKI models leverage a skeleton-language model, SkeletonCLIP, to infuse skeleton information into Vision Language Models (VLMs) and Large Vision Language Models (LVLMs) through collaborative training. Notably, SKI models do not require skeleton data during inference, enhancing their robustness for real-world applications. The effectiveness of SKI models is validated on three popular ADL datasets for zero-shot action recognition and video caption generation tasks.

Current Large Language Vision Models (LLVMs) trained on web videos perform well in general video understanding but struggle with fine-grained details, complex human-object interactions (HOI), and view-invariant representation learning essential for Activities of Daily Living (ADL). This limitation stems from a lack of specialized ADL video instruction-tuning datasets and insufficient modality integration to capture discriminative action representations. To address this, we propose a semi-automated framework for curating ADL datasets, creating ADL-X, a multiview, multimodal RGBS instruction-tuning dataset. Additionally, we introduce LLAVIDAL, an LLVM integrating videos, 3D skeletons, and HOIs to model ADL’s complex spatiotemporal relationships. For training LLAVIDAL a simple joint alignment of all modalities yields suboptimal results; thus, we propose a Multimodal Progressive (MMPro) training strategy, incorporating modalities in stages following a curriculum. We also establish ADL MCQ and video description benchmarks to assess LLVM performance in ADL tasks. Trained on ADL-X, LLAVIDAL achieves state-of-the-art performance across ADL benchmarks. Code and data will be made publicly available at https://adl-x.github.io/.

Human pose estimation (HPE) has received increasing attention recently due to its wide application in motion analysis, virtual reality, healthcare, etc. However, it suffers from the lack of labeled diverse real-world datasets due to the timeand labor-intensive annotation. To cope with the label deficiency issue, one common solution is to train the HPE models with easily available synthetic datasets (source) and apply them to real-world data (target) through domain adaptation (DA). Unfortunately, prevailing domain adaptation techniques within the HPE domain remain predominantly fixated on effecting alignment and aggregation between source and target features, often sidestepping the crucial task of excluding domain-specific representations. To rectify this, we introduce a novel framework that capitalizes on both representation aggregation and segregation for domain adaptive human pose estimation. Within this framework, we address the network architecture aspect by disentangling representations into distinct domain-invariant and domain-specific components, facilitating aggregation of domaininvariant features while simultaneously segregating domainspecific ones. Moreover, we tackle the discrepancy measurement facet by delving into various keypoint relationships and applying separate aggregation or segregation mechanisms to enhance alignment. Extensive experiments on various benchmarks, e.g., Human3.6M, LSP, H3D, and FreiHand, show that our method consistently achieves state-of-the-art performance.

Collaborative perception significantly enhances individual vehicle perception performance through the exchange of sensory information among agents. However, real-world deployment faces challenges due to bandwidth constraints and inevitable calibration errors during information exchange. To address these issues, we propose mmCooper, a novel multi-agent, multi-stage, communication-efficient, and collaboration-robust cooperative perception framework. Our framework leverages a multi-stage collaboration strategy that dynamically and adaptively balances intermediate- and late-stage information to share among agents, enhancing perceptual performance while maintaining communication efficiency. To support robust collaboration despite potential misalignments and calibration errors, our framework prevents misleading low-confidence sensing information from transmission and refines the received detection results from collaborators to improve accuracy. The extensive evaluation results on both real-world and simulated datasets demonstrate the effectiveness of the mmCooper framework and its components.

Recent advancements in 3D human pose estimation from single-camera images and videos have relied on parametric models, like SMPL. However, these models over-simplify anatomical structures, limiting their accuracy in capturing true joint locations and movements, which reduces their applicability in biomechanics, healthcare, and robotics. Biomechanically accurate pose estimation, on the other hand, typically requires costly marker-based motion capture systems and optimization techniques in specialized labs. To bridge this gap, we propose BioPose, a novel learning-based framework for predicting biomechanically accurate 3D human pose directly from monocular videos. BioPose includes three key components: a Multi-Query Human Mesh Recovery model (MQ-HMR), a Neural Inverse Kinematics (NeurIK) model, and a 2D-informed pose refinement technique. MQ-HMR leverages a multi-query deformable transformer to extract multi-scale fine-grained image features, enabling precise human mesh recovery. NeurIK treats the mesh vertices as virtual markers, applying a spatial-temporal network to regress biomechanically accurate 3D poses under anatomical constraints. To further improve 3D pose estimations, a 2D-informed refinement step optimizes the query tokens during inference by aligning the 3D structure with 2D pose observations. Experiments on benchmark datasets demonstrate that BioPose significantly outperforms state-of-the-art methods.

Human mesh recovery (HMR) provides rich human body information for various real-world applications such as gaming, human-computer interaction, and virtual reality. While image-based HMR methods have achieved impressive results, they often struggle to recover humans in dynamic scenarios, leading to temporal inconsistencies and non-smooth 3D motion predictions due to the absence of human motion. In contrast, video-based approaches leverage temporal information to mitigate this issue. In this paper, we present DiffMesh, an innovative motion-aware diffusion framework for video-based HMR. DiffMesh establishes a bridge between diffusion models and human motion, efficiently generating accurate and smooth output mesh sequences by incorporating human motion within the forward process and reverse process in the diffusion model. Extensive experiments are conducted on the widely used datasets (Human3.6M [15] and 3DPW [48]), which demonstrate the effectiveness and efficiency of our DiffMesh. Visual comparisons in real-world scenarios further highlight DiffMesh's suitability for practical applications. The project webpage is: https://zczcwh.github.io/diffmesh_page/

Human activity recognition (HAR) is crucial for real-world applications such as healthcare, surveillance, and smart homes. Among sensing technologies, millimeter wave (mmWave) sensors stand out due to their contactless nature, high sensitivity, and ability to operate in low-light environments while preserving privacy. However, the scarcity of mmWave sensing data limits the generalizability of mmWave-based HAR systems. To address this, we propose mmAP, a data augmentation and pretraining framework that synthesizes a large mmWave dataset using human mesh data, followed by pretraining a robust and general mmWave heatmap encoder using a multi-modal masked autoencoder framework using the synthesized data. We enhance the model’s robustness with heatmap-specific data perturbations and perform task-specific fine-tuning on a small real-world dataset. The experiment results over the baseline demonstrate the effectiveness of the proposed mmAP framework.

Federated learning (FL) has emerged as a promising paradigm for enabling the collaborative training of models without centralized access to the raw data on local devices. In the typical FL paradigm (e.g., FedAvg), model weights are sent to and from the server each round to participating clients. Recently, the use of small pre-trained models has been shown to be effective in federated learning optimization and improving convergence. However, recent state-of-the-art pre-trained models are getting more capable but also have more parameters, known as the "Foundation Models." In conventional FL, sharing the enormous model weights can quickly put a massive communication burden on the system, especially if more capable models are employed. Can we find a solution to enable those strong and readily available pre-trained models in FL to achieve excellent performance while simultaneously reducing the communication burden? To this end, we investigate the use of parameter-efficient fine-tuning in federated learning and thus introduce a new framework: FedPEFT. Specifically, we systemically evaluate the performance of FedPEFT across a variety of client stability, data distribution, and differential privacy settings. By only locally tuning and globally sharing a small portion of the model weights, significant reductions in the total communication overhead can be achieved while maintaining competitive or even better performance in a wide range of federated learning scenarios, providing insight into a new paradigm for practical and effective federated systems.

Recent advances in text-to-motion generation using dif-fusion and autoregressive models have shown promising re-sults. However, these models often suffer from a trade-off between real-time performance, high fidelity, and motion editability. To address this gap, we introduce MMM, a novel yet simple motion generation paradigm based on Masked Motion Model. MMM consists of two key components: (1) a motion tokenizer that transforms 3D human motion into a sequence of discrete tokens in latent space, and (2) a conditional masked motion transformer that learns to predict randomly masked motion tokens, conditioned on the pre-computed text tokens. By attending to motion and text to-kens in all directions, MMM explicitly captures inherent dependency among motion tokens and semantic mapping between motion and text tokens. During inference, this al-lows parallel and iterative decoding of multiple motion to-kens that are highly consistent with fine-grained text de-scriptions, therefore simultaneously achieving high-fidelity and high-speed motion generation. In addition, MMM has innate motion editability. By simply placing mask tokens in the place that needs editing, MMM automatically fills the gaps while guaranteeing smooth transitions between editing and non-editing parts. Extensive experiments on the HumanML3D and KIT-ML datasets demonstrate that MMM surpasses current leading methods in generating high-quality motion (evidenced by superior FID scores of 0.08 and 0.429), while offering advanced editing features such as body-part modification, motion in-betweening, and the synthesis of long motion sequences. In addition, MMM is two orders of magnitude faster on a single mid-range GPU than editable motion diffusion models. Our project page is available at https://exitudio.github.io/MMM-page/.

Most existing gait recognition methods are appearance-based, which rely on the silhouettes extracted from the video data of human walking activities. The less-investigated skeleton-based gait recognition methods directly learn the gait dynamics from 2D/3D human skeleton sequences, which are theoretically more robust solutions in the presence of appearance changes caused by clothes, hairstyles, and carrying objects. However, the performance of skeleton-based solutions is still largely behind the appearance-based ones. This paper aims to close such performance gap by proposing a novel network model, GaitMixer, to learn more discriminative gait representation from skeleton sequence data. In particular, GaitMixer follows a heterogeneous multi-axial mixer architecture, which exploits the spatial self-attention mixer followed by the temporal large-kernel convolution mixer to learn rich multi-frequency signals in the gait feature maps. Experiments on the widely used gait database, CASIA-B, demonstrate that GaitMixer outperforms the previous SOTA skeleton-based methods by a large margin while achieving a competitive performance compared with the representative appearance-based solutions. Code will be available at https://github.com/exitudio/gaitmixer

Most existing deep neural networks are static, which means they can only perform inference at a fixed complexity. But the resource budget can vary substantially across different devices. Even on a single device, the affordable budget can change with different scenarios, and repeatedly training networks for each required budget would be incredibly expensive. Therefore, in this work, we propose a general method called MutualNet to train a single network that can run at a diverse set of resource constraints. Our method trains a cohort of model configurations with various network widths and input resolutions. This mutual learning scheme not only allows the model to run at different width-resolution configurations but also transfers the unique knowledge among these configurations, helping the model to learn stronger representations overall. MutualNet is a general training methodology that can be applied to various network structures (e.g., 2D networks: MobileNets, ResNet, 3D networks: SlowFast, X3D) and various tasks (e.g., image classification, object detection, segmentation, and action recognition), and is demonstrated to achieve consistent improvements on a variety of datasets. Since we only train the model once, it also greatly reduces the training cost compared to independently training several models. Surprisingly, MutualNet can also be used to significantly boost the performance of a single network, if dynamic resource constraints are not a concern. In summary, MutualNet is a unified method for both static and adaptive, 2D and 3D networks. Code and pre-trained models are available at https://github.com/taoyang1122/MutualNet.

The UnderWater Acoustic (UWA) communication is the foundation for oceanic information applications. However, existing UWA systems suffer from low data rate problem. Although the Multiple Input Multiple Output (MIMO) scheme is proved to be able to increase channel capacity in many terrestrial scenarios, the high cost and complexity of acoustic MIMO significantly limit its usage. Recently, the acoustic reconfigurable intelligent surfaces (acoustic RIS) concept has been proposed to address the aforementioned problem. In this paper, with the real-world constraints taken into consideration, three key components of the acoustic RIS are designed to realize the underwater RIS concept, including the new acoustic RIS hardware, the ultra-wideband (UWB) beamforming, and practical operation protocol. Specifically, a completely new hardware design of acoustic RIS is first provided, since existing electromagnetic RIS designed for terrestrial environments do not work for underwater acoustic waves. Then, the UWB beamforming solution is developed, since underwater acoustic RIS need to control the acoustic signals, whose bandwidth is comparable to its carrier frequency. Finally, the practical operation protocol is developed to realize the acoustic RIS functionalities in complex underwater environment. The acoustic RIS design is validated through both COMSOL multiphysics simulations and end-to-end Bellhop-based simulations.

Handwritten signature authentication is an important service to defend against fraudulent activities. Current automated solutions rely heavily on dedicated devices and require certain user efforts. In this work, we explore the feasibility of a new type of signature authentication system, SAP - Signature Authentication with PPG Sensor, which leverages Photoplethysmography (PPG) sensors in wrist-worn wearable devices. To make SAP non-intrusive and secure, we design effective algorithms to separate the signature signals from the heartbeat signals in the raw PPG signals. We implement a low-cost hardware prototype of SAP. Our preliminary experimental results show that SAP can achieve an average F1 score of up to 98%.

Federated learning (FL) is a promising strategy for performing privacy-preserving, distributed learning with a network of clients (i.e., edge devices). However, the data distribution among clients is often non-IID in nature, making efficient optimization difficult. To alleviate this issue, many FL algorithms focus on mitigating the effects of data heterogeneity across clients by introducing a variety of proximal terms, some incurring considerable compute and/or memory overheads, to restrain local updates with respect to the global model. Instead, we consider rethinking solutions to data heterogeneity in FL with a focus on local learning generality rather than proximal restriction. To this end, we first present a systematic study informed by second-order indicators to better understand algorithm effectiveness in FL. Interestingly, we find that standard regularization methods are surprisingly strong performers in mitigating data heterogeneity effects. Based on our findings, we further propose a simple and effective method, FedAlign, to overcome data heterogeneity and the pitfalls of previous methods. FedAlign achieves competitive accuracy with state-of-the-art FL methods across a variety of settings while minimizing computation and memory overhead. Code is available at https://github.com/mmendiet/FedAlign.

Physical activity (PA) plays a crucial role in leading a healthy life without chronic diseases. Among various PAs, weight lifting, one of the essential stationary exercises, is an integral part of routine workout sessions. Being aware of the intensity of the performed exercise is also an essential factor in keeping track of the workout. Inspired by this, we propose a low-cost quantitative weight lifting assessment system, PaWLA, leveraging only a single Photoplethysmography (PPG) sensor. Particularly, we design PaWLA as a mobile weight recognition system that can classify the user’s lifted weight into its corresponding label based on PPG sensor readings from the wrist region. The changes in blood volume in the radial artery due to the strain of lifting the weight are exploited via PPG sensor readings in this work. We build our custom hardware prototype using COTS components to prove the system’s feasibility. Evaluation of the system with nine volunteers shows that PaWLA can achieve an average F1 score of up to 97.4%, proving the feasibility and efficiency of the proposed method.

Existing deep learning-based human mesh reconstruction approaches have a tendency to build larger networks to achieve higher accuracy. Computational complexity and model size are often neglected, despite being key characteristics for practical use of human mesh reconstruction models (e.g. virtual try-on systems). In this paper, we present GTRS, a lightweight pose-based method that can reconstruct human mesh from 2D human pose. We propose a pose analysis module that uses graph transformers to exploit structured and implicit joint correlations, and a mesh regression module that combines the extracted pose feature with the mesh template to reconstruct the final human mesh. We demonstrate the efficiency and generalization of GTRS by extensive evaluations on the Human3.6M and 3DPW datasets. In particular, GTRS achieves better accuracy than the SOTA pose-based method Pose2Mesh while only using 10.2% of the parameters (Params) and 2.5% of the FLOPs on the challenging in-the-wild 3DPW dataset. Code is available at https://github.com/zczcwh/GTRS

Handwritten signature authentication is a crucial service to defend against fraudulent activities. Existing automated solutions rely heavily on dedicated devices that are expensive and require different user efforts that affect the user experience. In this paper, we propose a new signature authentication system, PPGSign, which leverages Photoplethysmography (PPG) sensors in the existing wrist-worn wearable devices. The unique blood flow changes in the supplicant’s hand movement are exploited in this system to validate the signature. To make PPGSign nonintrusive and secure, we explore effective algorithms to separate the signature signals from the heartbeat signals in the raw PPG signals. We build a low-cost hardware prototype to verify our proposed method. Our experimental results show that PPGSign can achieve an average F1 score of up to 98%, which verifies the feasibility and efficiency of the proposed solution.

With the wide bandwidths in millimeter wave (mmWave) frequency band that results in unprecedented accuracy, mmWave sensing has become vital for many applications, especially in autonomous vehicles (AVs). In addition, mmWave sensing has superior reliability compared to other sensing counterparts such as camera and LiDAR, which is essential for safety-critical driving. Therefore, it is critical to understand the security vulnerabilities and improve the security and reliability of mmWave sensing in AVs. To this end, we perform the end-to-end security analysis of a mmWave-based sensing system in AVs, by designing and implementing practical physical layer attack and defense strategies in a state-of-the-art mmWave testbed and an AV testbed in real-world settings. Various strategies are developed to take control of the victim AV by spoofing its mmWave sensing module, including adding fake obstacles at arbitrary locations and faking the locations of existing obstacles. Five real-world attack scenarios are constructed to spoof the victim AV and force it to make dangerous driving decisions leading to a fatal crash. Field experiments are conducted to study the impact of the various attack scenarios using a Lincoln MKZ-based AV testbed, which validate that the attacker can indeed assume control of the victim AV to compromise its security and safety. To defend the attacks, we design and implement a challenge-response authentication scheme and a RF fingerprinting scheme to reliably detect aforementioned spoofing attacks.

The millimeter wave (mmWave) technologies can significantly increase the throughput and user capacity in the future wireless networks. In term of device authentication, due to the usage of highly directional communication link, new physical layer identification (PLI) mechanism based on the spatial-temporal beam features becomes available. However, it is not known how to implement the new PLI mechanism using commodity devices in multiple client scenario in wireless networks. To this end, this paper presents a practical operational framework for the new beam feature-based PLI that is compatible with 802.11ad/ay standards. The low cost of these commodity devices leads to much wider beams, multiple main lobes, and high side lobes which in turn results in frequent sector level sweep (SLS) even for a minimal level of the transmitter-receiver misalignment. The high mobility sensitivity also triggers SLS. The key idea is to utilize the mobility of the mmWave device to collect enough measurements, the beam pattern feature values, from different observation angles where the beam features are extracted. This mobility effect takes advantage of the rich spatial-temporal information of the feature to prevent the system from spoofing. We also propose a novel feature database refinement algorithm to strengthen the database against false accept/reject rates and increase the identification accuracy. The algorithm filters the noisy data collected in the presence of multiple-clients. The proposed operational framework is implemented in commodity 802.11ad/ay devices. We show that the proposed scheme can reach near 100% accuracy even with a minimal feature vector database in real-time scenarios.

With millimeter wave (mmWave) wireless communication envisioned to be the key enabler of next generation high data rate wireless networks, security is of paramount importance. While conventional security measures in wireless networks operate at a higher layer of the protocol stack, physical layer security utilizes unique device dependent hardware features to identify and authenticate legitimate devices. In this work, we identify that the manufacturing tolerances in the antenna arrays used in mmWave devices contribute to a beam pattern that is unique to each device, and to that end we propose a novel device fingerprinting scheme based on the unique beam pattern of different codebooks used by the mmWave devices. Specifically, we propose a fingerprinting scheme with multiple access points (APs) to take advantage of the rich spatial-temporal information of the beam pattern. We perform comprehensive experiments with commercial off-the-shelf mmWave devices to validate the reliability performance of our proposed method under various scenarios. We also compare our beam pattern feature with a conventional physical layer feature namely power spectral density feature (PSD). To that end, we implement PSD feature based fingerprinting for mmWave devices. We show that the proposed multiple APs scheme is able to achieve over 99% identification accuracy for stationary LOS and NLOS scenarios and significantly outperform the PSD feature fingerprinting method. For mobility scenario, the overall identification accuracy is 99%. In addition, we perform security analysis of our proposed beam pattern fingerprinting system and PSD fingerprinting system by studying the feasibility of performing impersonation attacks. We design and implement an impersonation attack mechanism for mmWave wireless networks using state-of-the-art 60 GHz software defined radios. We discuss our findings and their implications on the security of the mmWave wireless networks.

Multi-hop networks (e.g., mesh, ad-hoc, and sensor networks) are important and cost-efficient communication backbones. Over the last few years wireless data traffic has drastically increased due to the changes in the way today's society creates, shares, and consumes information. This demands the efficient and intelligent utilization of limited network resources to optimize network performance. Traffic engineering (TE) optimizes network performance and enables optimal forwarding and routing rules to meet the quality of service (QoS) requirements for a large volume of traffic flows. This paper proposes a distributed model-free TE solution based on stochastic policy gradient reinforcement learning (RL), which aims to learn a stochastic routing policy for each router so that each router can send a packet to the next-hop router according to the learned optimal probability. The proposed policy-gradient solution naturally leads to multi-path TE strategies, which can effectively distribute the high traffic loads among all available routing paths to minimize the E2E delay. Moreover, a distributed software-defined networking architecture is proposed, which enables the fast prototyping of the proposed multi-agent actor-critic TE (MA-AC TE) algorithm and in-nature supports automated TE through multi-agent RL learning.

Underwater magnetic induction communication (UWMIC) provides a low-power and high-throughput solution for autonomous underwater vehicles (AUVs). UWMIC with tri-axis coils increases the reliability of wireless channel by exploring the coil orientation diversity. However, the UWMIC channel is different from typical fading channels and the mutual inductance information (MII) is not always available. It is not clear the performance of the tri-axis coil MIMO without MII. Also, its performances with multiple users have not been investigated. In this paper, we analyze the reliability and multiplexing gain of UWMIC with tri-axis coils by using coil selection. We optimally select the transmit and receive coils to reduce the computation complexity and power consumption and explore the diversity for multiple users. We find that without using all the coils and MII, we can still achieve reliability. Also, the multiplexing gain of UWMIC without MII is 5 dB smaller than typical terrestrial fading channels. The results of this paper provide a more power-efficient way to use UWMIC with tri-axis coils.

The acoustic cooperative multi-input-multi-output (MIMO) systems equipped on the underwater robot swarms (URSs) can enable long-range and high-throughput communications. However, the acoustic communications cannot provide the real-time and accurate synchronization for the distributed transmitters of the cooperative MIMO due to the large delay of acoustic channels. In addition, the narrow bandwidth of the acoustic channel further enlarges the synchronization time and errors. In this paper, we propose the metamaterial magnetic induction (M2I)-assisted acoustic cooperative MIMO to address aforementioned challenges. The synchronization time can be reduced since the M2I has negligible signal propagation delays. To quantitatively analyze the improvement, we deduce the synchronization errors, signal-to-noise ratio (SNR), effective communication time, and the throughput of the system. Finally, the improvement of using M2I-assisted synchronization is validated by the numerical evaluation.

Next generation wireless communication networks utilizing 60 GHz millimeter wave (mmWave) frequency bands are expected to achieve multi-gigabit throughput with the use of highly directional phased-array antennas. These directional signal beams provide enhanced security to the legitimate networks due to the increased difficulties of eavesdropping. However, there still exists significant possibility of eavesdropping since (i) the reflections of the signal beam from ambient reflectors enables opportunistic stationary eavesdropping attacks; and (ii) carefully designed beam exploration strategy enables active nomadic eavesdropping attack. This paper discusses eavesdropper attack strategies for 802.11ad mmWave systems and provides the first analytical model to characterize the success possibility of eavesdropping in both opportunistic stationary attacks and active nomadic attacks.

This paper proposes NeuralWave, an intelligent and non-intrusive user identification system based on human gait biometrics extracted from WiFi signals. In particular, the channel state information (CSI)measurements are first collected from commodity WiFi devices. Then, a collection of data preprocessing schemes are applied to sanitize and calibrate the noisy and erroneous CSI data samples to manifest and augment the gait-induced radio-frequency (RF)signatures. Next, a 23-layer deep convolutional neural network, namely RadioNet, is developed to automatically learn the salient features from the preprocessed CSI data samples. The extracted features constitute a latent representation for the gait biometric that is discriminative enough to distinguish one person from another. Using the latent biometric representation, a softmax multi-class classifier is adopted to achieve accurate user identification. Extensive experiments in a typical indoor environment are conducted to show the effectiveness of our system. In particular, NeuralWave can achieve 87.76 ± 2.14% user identification accuracy for a group of 24 people. To the best of our knowledge, NeuralWave is the first in the literature to exploit deep learning for feature extraction and classification of physiological and behavioral gait biometrics embedded in CSI signals from commodity WiFi.

Supporting on-line and adaptive traffic engineering in software-defined networks entails the fast, robust control message forwarding from software-defined switches to the controller(s). In-band control using the existing infrastructure is cost-efficient, but imposes a substantial barrier to timely transmissions of control messages. Also, due to the limited computational capability of a single controller, only the use of multiple controllers is practically viable for large-scaled networks. Therefore, in this paper, the optimal software-defined network planning is investigated with multi-controllers. First, the network planning problem is formulated as a nonlinear multi-objective optimization, which aims to simultaneously minimize the number of controllers and the control traffic delay for each switch. This planning problem is then partitioned into two sub-problems, i.e., multi-controller placement and control traffic balancing, which are respectively solved by the proposed fast-convergent algorithms. Furthermore, an adaptive feedback control mechanism is proposed to iteratively work out the two sub-problems and enable the dynamic network replanning, subject to the time-varying traffic volume and network topology. Simulations validate the adaptivity of our control scheme, which significantly reduces delay with maximum throughput for control flows, brings minimal impact to normal data flows, and requires the minimum controllers.

Providing high-speed communication for mobile users in remote geographic areas or after a disaster occurs is not only critical but also challenging. To counter such challenge, unmanned aerial vehicles (UAVs) have been exploited to provide a fast-deployable and high-speed communication system, where each UAV can serve as an aerial small cell base station to provide WiFi and/or cellular services for the ground users. Despite its fast-deployable and highly maneuverable features, the capacity analysis of aerial small cells is largely missing. To close such gap, a stochastic propagation model for A-to-G aerial channels is first introduced, which takes into account the impact from wave propagation, gaseous absorption, Doppler spread, attitude-dependent shadowing, and multipath fading. Then, by exploiting such model, the area spectral efficiency of the aerial small cells is investigated for both SISO and MIMO cases. Our study reveals the inherent relationship among the area capacity, height and coverage and shows that there exists an optimal attitude that can maximize network capacity and cell coverage.

Recent years have witnessed an active research on the stochastic network utility maximization (NUM) problem, where the optimal network resource allocation policy, such as congestion control, routing, and scheduling, is formulated as a constrained maximization of some utility function under the stochastic dynamics in users' traffic and time-varying wireless channels. So far, the stochastic NUM problem and its associated solutions are investigated under light-tailed (LT) traffic assumption, which is largely departing from the recent large-scale empirical studies, which verify the emergence of heavy-tailed (HT) traffic in a variety of networked systems. The unique stochastic features of HT traffic fundamentally challenges the feasibility of the classic algorithms developed for the stochastic NUM problem. This paper aims to develop effective algorithms to maximize network utility in the presence of heavy tailed traffic. First, it is proven that without considering the inherent bursty nature of HT traffic, the classic stochastic subgradient algorithm, which is utility optimal, fail to achieve queue stability. Our theoretical analysis reveals that such queue instability phenomenon is due to the fact that the stochastic subgradient algorithm inherently exploits queue length to update Lagrange dual variables. To counter this challenge, a time-average stochastic gradient algorithm is proposed, which decouples the queue-length update process and the dual variable update process in such a way that utility-optimality and queue-stability can be simultaneously achieved. Our convergence and stability analysis shows that the proposed algorithm can avoid the LT queues from competing with the HT queues, thus completely shielding those LT queues from the destructive impact of HT traffic.

Wireless underground sensor networks enable many applications, such as mine and tunnel disaster prevention, oil upstream monitoring, earthquake prediction and landslide detection, and intelligent farming and irrigation among many others. Most applications are location-dependent, so they require precise sensor positions. However, classical localization solutions based on the propagation properties of electromagnetic waves do not function well in underground environments. This paper proposes a magnetic induction (MI)-based localization that accurately and efficiently locates randomly deployed sensors in underground environments by leveraging the multipath fading free nature of MI signals. Specifically, the MI-based localization framework is first proposed based on underground MI channel modeling with additive white Gaussian noise, the designated error function, and semidefinite programming relaxation. Next, this paper proposes a two-step positioning mechanism for obtaining fast and accurate localization results by: first, developing the fast-initial positioning through an alternating direction augmented Lagrangian method for rough sensor locations within a short processing time, and then proposing fine-grained positioning for performing powerful search for optimal location estimations via the conjugate gradient algorithm. Simulations confirm that our solution yields accurate sensor locations with both low and high noise and reveals the fundamental impact of underground environments on the localization performance.

Heavy-tailed (HT) traffic (e.g., the Internet and multimedia traffic) fundamentally challenges the validity of classic scheduling algorithms, designed under conventional light-tailed (LT) assumptions. To address such a challenge, this paper investigates the impact of HT traffic on delay-based maximum weight scheduling (DMWS) algorithms, which have been proven to be throughput-optimal with enhanced delay performance under the LT traffic assumption. First, it is proven that the DMWS policy is not throughput-optimal anymore in the presence of hybrid LT and HT traffic by inducing unbounded queuing delay for LT traffic. Then, to solve the unbounded delay problem, a delay-based maximum power-weight scheduling (DMPWS) policy is proposed that makes scheduling decisions based on queuing delay raised to a certain power. It is shown by the fluid model analysis that DMPWS is throughput-optimal with respect to moment stability by admitting the largest set of traffic rates supportable by the network, while guaranteeing bounded queuing delay for LT traffic. Moreover, a variant of the DMPWS algorithm, namely the IU-DMPWS policy, is proposed, which operates with infrequent queue state updates. It is also shown that compared with DMPWS, the IU-DMPWS policy preserves the throughput optimality with much less signaling overhead, thus expediting its practical implementation.

Scheduling is one of the most important resource allocation for networked systems. Conventional scheduling policies are primarily developed under light-tailed (LT) traffic assumptions. However, recent empirical studies show that heavy-tailed (HT) traffic flows have emerged in a variety of networked systems, such as cellular networks, the Internet, and data centers. The highly bursty nature of HT traffic fundamentally challenges the applicability of the conventional scheduling policies. This paper aims to develop novel throughput-optimal scheduling algorithms under hybrid HT and LT traffic flows, where classic optimal policies (e.g., maximum-weight/backpressure schemes), developed under LT assumption, are not throughput-optimal anymore. To counter this problem, a delay-based maximum-weight scheduling policy with the last-in first-out (LIFO) service discipline, namely LIFO-DMWS, is proposed with the proved throughput optimality under hybrid HT and LT traffic. The throughput optimality of LIFO-DMWS gives that a networked system can support the largest set of incoming traffic flows, while guaranteeing bounded queueing delay to each queue, no matter the queue has HT or LT traffic arrival. Specifically, by exploiting asymptotic queueing analysis, LIFO-DMWS is proved to achieve throughout optimality without requiring any knowledge of traffic statistic information (e.g., the tailness or burstiness of traffic flows). Simulation results validate the derived theories and confirm that LIFO-DMWS achieves bounded delay for all flows under challenging HT environments.

Recently, large-scale spectrum measurements show that geo-location spectrum databases, as recommended by regulators (e.g., FCC, Ofcom, ECC) for TV white space (TVWS) discovery are notoriously inaccurate in Metropolitan areas because of inaccurate TV channel propagation models they adopted. To counter this challenge, we propose a cloud-based crowd-augmented spectrum mapping scheme. Our scheme aims to build accurate geo-location database in Metropolitan areas with high spatial resolution under minimum cost by jointly utilizing superior computing capacity of cloud servers, abundant spectrum sensing data from crowd of mobile white-spaces device (WSD) users and the well-established geo-statistical techniques. More specifically, our spectrum mapping scheme consists of three interdependent components (1) opportunistic mobile spectrum sensing, which exploits the high spatial diversity of mobile users along with low-cost embedded spectrum measuring solution to retrieve power spectrum density (PSD) information of the TV channels in a large Metropolitan region; (2) cloud-based geo-statistical spectrum mapping, which estimates PSD at unknown geographic locations by utilizing the abundant PSD data aggregated at the cloud server consisting of geo-statistical analysis and interpolation tools; (3) optimal spatial sampling, which further augments the accuracy of the spectrum map by selecting the optimal locations, at which additional spectrum sensing measurements are obtained to minimize the spectrum mapping error. To verify the performance of the proposed scheme, an experiment is conducted in our university campus. The experiment result shows that our proposed scheme can discover more spectrum opportunities than the information reported by commercially available geo-location spectrum databases.

Nanotechnology is enabling the development of miniature devices able to perform simple tasks at the nanoscale. The interconnection of such nano-devices with traditional wireless networks and ultimately the Internet enables a new networking paradigm known as the Internet of Nano-Things (IoNT). Despite their promising applications, nano-devices have constrained power, energy, and computation capabilities along with very limited memory on board, which may only be able to hold one packet at once and, thus, requires packets to be delivered before certain hard deadlines. Toward this goal, a fully-distributed computation-light provably-correct scheduling/MAC protocol is introduced for bufferless nano-devices, which can maximize the network throughput, while achieving perpetual operation. More specifically, the proposed scheduling algorithm allows every nano-device to make optimal transmission decisions locally based on its incoming traffic rate, virtual debts, and channel sensing results. It is proven that the proposed algorithm is timely throughput optimal in the sense that it can guarantee reliable data delivery before deadlines as long as the incoming traffic rates are within the derived maximum network capacity region. This feature not only can lead to high network throughput for the IoNT, but also guarantees that the memory of each device is empty before the next packet arrives, thus addressing the fundamental challenge imposed by the extremely limited memory of nano-devices. In addition, the optimal deadline is derived, which guarantees that all the nano-devices can achieve perpetual communications by jointly considering the energy consumption of communications over the terahertz channel and energy harvesting based on piezoelectric nano-generators.

While underwater wireless communications have been investigated and implemented for decades, existing solutions still have difficulties in establishing reliable and low-delay wireless underwater links among small-size devices. The Magnetic Induction (MI) communication technique is among the promising solutions due to its advantages in low propagation delay and less susceptibility to the transmission environments. To date, existing MI models cannot accurately characterize the complex underwater MI channels, especially in the shallow water with omnidirectional antennas. In this paper, an analytical channel model is developed for underwater MI communication system with Tri-directional coil (TD coil), which is derived based on the rigorous electromagnetic field analysis. The MIMO channel between the tri-directional coil antennas are characterized under the complex influences from the water absorption as well as the surface reflection and lateral waves. To validate the channel model, we compare the theoretical results with simulations derived by the COMSOL Multiphysics simulation tool. The developed channel model confirms the feasibility and lays the foundation of reliable MI underwater communications.

The wireless sensor networks can enable the real-time monitoring in the pipeline environments, which can facilitate important applications such as structure identification and fault diagnosis. The magnetic induction (MI)-based techniques provide efficient and reliable wireless communications among sensor nodes in such challenging environments. This paper proposes a localization strategy for MI-based wireless sensor network in the complex pipeline environment without requiring any additional infrastructure. The proposed localization strategy analytically captures the unique pipeline system effects, including antenna orientations, absorption of fluid medium, reflections on pipe walls, node's position on the pipe cross section, and the pipe turns and branches. The numerical evaluation shows the proposed strategy can provide high resolution position estimation for small wireless mobile sensor nodes with arbitrary orientation and position inside the pipelines.

The heavy tailed traffic in wireless networks fundamentally challenges the applicability of conventional throughput optimal scheduling algorithms. To encounter this, the stability performance of distributed maximum weight scheduling algorithms (DMWS), which are known to be throughput optimal under light tailed environment, is first analyzed.More specifically, it is shown that heavy tailed traffic can significantly degrade the stability performance of DMWS. In particular, it is proven that if a user with light-tailed traffic arrivals has the average traffic rate λ below a threshold λ*, it will experience bounded queueing delay. Otherwise, if λ is larger than a threshold λ', its queueing delay is necessarily of infinite mean. To address this problem, the distributed maximum weight-α scheduling (DMWS-α) algorithm is proposed, which makes the scheduling decision based on the queue lengths raised to the α-th power. It is demonstrated that DMWS-α is throughput optimal with respect to moment stability in the sense that if the traffic arrivals rates are within the network stability region, all network users with light-tailed traffic arrivals always have bounded queueing delay with finite mean and variance.

Wireless underground sensor networks (WUSNs) enable many applications such as underground pipeline monitoring, power grid maintenance, mine disaster prevention, and oil upstream monitoring among many others. While the classical electromagnetic waves do not work well in WUSNs, the magnetic induction (MI) propagation technique provides constant channel conditions via small size of antenna coils in the underground environments. In this paper, instead of adopting currently layered protocols approach, a distributed cross-layer protocol design is proposed for MI-based WUSNs. First, a detailed overview is given for different communication functionalities from physical to network layers as well as the QoS requirements of applications. Utilizing the interactions of different layer functionalities, a distributed environment-aware protocol, called DEAP, is then developed to satisfy statistical QoS guarantees and achieve both optimal energy savings and throughput gain concurrently. Simulations confirm that the proposed cross-layer protocol achieves significant energy savings, high throughput efficiency and dependable MI communication for WUSNs.

Click on the DOI link to access the article (may not be free).

Recent advance in distributed scheduling algorithms mainly focuses on designing CSMA-type protocols to achieve maximum network throughput in a fully distributive manner. However, it is inherently difficult for distributed scheduling algorithms to promise high throughput while maintaining low delay. To encounter this, a new scheduling algorithm, namely timely-throughput optimal algorithm, is introduced, which distributively determines the optimal transmission times for network users so that the largest set of traffic rates of network users can be supported, while ensuring timely data delivery within hard deadlines. More specifically, the maximum network capacity region that timely-throughput optimal scheduling algorithms can achieve is first derived, which characterizes the closure of the set of all arrival rate vectors under which there exists an viable scheduling algorithm to guarantee that all network users can meet the delay deadlines. Then, a distributed scheduling algorithm is proposed, which allows every user to make optimal transmission decision locally based on its incoming traffic rate, virtual debts, and previous channel sensing results. Finally, it is proven that the proposed algorithm is timely-throughput optimal in the sense that it can guarantee reliable data delivery before deadlines as long as the incoming traffic rates are within the derived maximum network capacity region.

The heavy tailed (HT) traffic from wireless users, caused by the emerging Internet and multimedia applications, introduces a HT interference region within which network users will experience unbounded delay with infinite mean and/or variance. Specifically, it is proven that, if the network traffic of primary networks (e.g., cellular and Wi-Fi networks) is heavy tail distributed, there always exists a critical density λpsuch that, if the density of primary users is larger than λp, the secondary network users (e.g., sensor devices and cognitive radio users) can experience unbounded end-to-end delay with infinite variance even though there exists feasible routing paths along the network users. To counter this problem, the mobility of network users is utilized to achieve delay-bounded connectivity, which simultaneously ensures the existence of routing paths and the finiteness of the delay variance along these paths. In particular, it is shown that there exists a critical threshold on the maximum radius that the secondary user can reach, above which delay-bounded connectivity is achievable in the secondary networks. In this case, the end-to-end latency of secondary users is shown to be asymptotically linear in the Euclidean distance between the transmitter and receiver.

The heavy tailed nature exhibited in both primary and secondary users' traffic fundamentally challenges the performance limit of dynamic spectrum access (DSA) networks under the conventional light tailed assumptions. This paper provides an asymptotic analysis of the steady-state queue length distribution of secondary users (SUs) under the heavy tailed network environment. Specifically, two network scenarios are investigated. In the first scenario where each SU has its exclusive access to a primary user (PU) channel, it is shown that the heavy tailed nature of either the PU traffic or the SU traffic can make SUs experience heavy tailed queue length with unbounded moments. In the second scenario where multiple SUs share a single PU channel, the queuing performance under throughput optimal scheduling policies is studied. It is proven that if the PU traffic has a heavier tail than any SU traffic, the queue length of each SU is at least one order heavier than the PU traffic under any scheduling policy. Otherwise, if the traffic from at least one of the SUs has a heavier tail than the PU traffic, it is proven that the celebrated throughput-optimal maximum weight scheduling leads to the worst possible asymptotic queuing performance for SUs by letting each SU queue have the heaviest possible tail. On the contrary, it is shown that there always exists a feasible set of β parameters such that the maximum weight-β scheduling yields the best asymptotic performance for the SU queues by letting each queue have the lightest possible tail.

In wireless multimedia sensor networks (WMSNs), visual correlation exists among multiple nearby cameras, thus leading to considerable redundancy in the collected images. This paper proposes a differential coding-based scheduling framework for efficiently gathering visually correlated images. This framework consists of two components including MinMax Degree Hub Location (MDHL) and Maximum Lifetime Scheduling (MLS). The MDHL problem aims to find the optimal locations for the multimedia processing hubs, which operate on different channels for concurrently collecting images from adjacent cameras, such that the number of channels required for frequency reuse is minimized. After associating camera sensors with proper hubs, the MLS problem targets at designing a schedule for the cameras such that the network lifetime of the cameras is maximized by letting highly correlated cameras perform differential coding on the fly. It is proven in this paper that the MDHL problem is NP-complete, and the MLS problem is NP-hard. Consequently, approximation algorithms are proposed to provide bounded performance. Since the designed algorithms only take the camera settings as inputs, they are independent of specific multimedia applications. Experiments and simulations show that the proposed differential coding-based scheduling can effectively enhance the network throughput and the energy efficiency of camera sensors.

The heavy tailed nature in dynamic spectrum networks challenges the applicability of conventional network stability criterions. To encounter this, a new stability criterion, namely moment stability, is introduced, which requires that the queue length of each secondary user has finite moments for every achievable order. Then, the necessary and sufficient conditions for the existence of a resource allocation policy to achieve moment stability are derived. Moreover, the network stability region yielded from these conditions is shown to be directly related to the statistics of secondary user traffics, primary user activities, the number of secondary users contending the spectrum, and the total number of primary user channels available to secondary users. In addition, a throughput-optimal policy, which stabilizes the network for any arrival rates in the stability region, is also introduced. In the end, the impact of the heavy tailed primary user traffic on the network stability is investigated, which shows that the tail heaviness of the primary user traffics determines what types of the network stability are achievable for a dynamic spectrum networks.

This paper provides an asymptotic analysis of the transmission delay experienced by SUs for dynamic spectrum access (DSA) networks. It is shown that DSA induces only light-tailed delay if both the busy time of PU channels and the message size of SUs are light tailed. On the contrary, if either the busy time or the message size is heavy tailed, then the SUs' transmission delay is heavy tailed. For this latter case, it is proven that if one of either the busy time or the message size is light tailed and the other is regularly varying with index \alpha, the transmission delay is regularly varying with the same index. As a consequence, the delay has an infinite mean provided \alpha < 1 and an infinite variance provided \alpha < 2. Furthermore, if both the busy time and the message size are regularly varying with different indices, then the delay tail distribution is as heavy as the one with the smaller index. Moreover, the impact of spectrum mobility and multiradio diversity on the delay performance of SUs is studied. It is shown that both spectrum mobility and multiradio diversity can greatly mitigate the heavy-tailed delay by increasing the orders of its finite moments.

The spatial correlation of visual information retrieved from distributed camera sensors leads to considerable data redundancy in wireless video sensor networks, resulting in significant performance degradation in energy efficiency and quality-of-service (QoS) satisfaction. In this paper, a correlation-aware QoS routing algorithm (CAQR) is proposed to efficiently deliver visual information under QoS constraints by exploiting the correlation of visual information observed by different camera sensors. First, a correlation-aware inter-node differential coding scheme is designed to reduce the amount of traffic in the network. Then, a correlation-aware load balancing scheme is proposed to prevent network congestion by splitting the correlated flows that cannot be reduced to different paths. Finally, the correlation-aware schemes are integrated into an optimization QoS routing framework with an objective to minimize energy consumption subject to delay and reliability constraints. Simulation results demonstrate that the proposed routing algorithm achieves efficient delivery of visual information under QoS constraints in wireless video sensor networks.

Sandstorms are serious natural disasters, which are commonly seen in the Middle East, Northern Africa, and Northern China.In these regions, sandstorms have caused massive damages to the natural environment, national economy, and human health. To avoid such damages, it is necessary to effectively monitor the origin and development of sandstorms. To this end, wireless sensor networks (WSNs) can be deployed in the regions where sandstorms generally originate so that sensor nodes can collaboratively perform sandstorm monitoring and rapidly convey the observations to remote administration center. Despite the potential advantages, the deployment of WSNs in the vicinity of sandstorms faces many unique challenges, such as the temporally buried sensors and increased path loss during sandstorms. Consequently, the WSNs may experience frequent disconnections during the sandstorms. This further leads to dynamically changing topology. In this paper, a topology analysis of the WSNs for sandstorm monitoring is performed. Four types of channels a sensor can utilize during sandstorms are analyzed, which include air-to-air channel, air-to-sand channel, sand-to-air channel, and sand-to-sand channel. Based on the channel model solutions, a percolation-based connectivity analysis is performed. It is shown that if the sensors are buried in low depth, allowing sensor to use multiple types of channels improves network connectivity. Accordingly, much smaller sensor density is required compared to the case, where only terrestrial air channels are used. Through this topology analysis a WSN architecture can be deployed for very efficient sandstorm monitoring.

In wireless multimedia sensor networks (WMSNs), visual correlation exist among multiple nearby cameras, thus leading to considerable redundancy in the collected images. This paper addresses the problem of timely and efficiently gathering visually correlated images from camera sensors. Towards this, three fundamental problems are considered, namely, MinMax Degree Hub Location (MDHL), Minimum Sum-entropy Camera Assignment (MSCA), and Maximum Lifetime Scheduling (MLS). The MDHL problem aims to find the optimal locations to place the multimedia processing hubs, which operate on different channels for concurrently collecting images from adjacent cameras, such that the number of channels required for frequency reuse is minimized. With the locations of the hubs determined by the MDHL problem, the objective of the MSCA problem is to assign each camera to a hub in such a way that the global compression gain is maximized by jointly encoding the visually correlated images gathered by each hub. At last, given a hub and its associated cameras, the MLS problem targets at designing a schedule for the cameras such that the network lifetime of the cameras is maximized by letting highly correlated cameras perform differential coding on the fly. It is proven in this paper that the MDHL problem is NP-complete, and the others are NP-hard. Consequently, approximation and heuristic algorithms are proposed. Since the designed algorithms only take the camera settings as inputs, they are independent of specific multimedia applications. Experiments and simulations show that the proposed image gathering schemes effectively enhance network throughput and image compression performance.

Data redundancy caused by correlation has motivated the application of collaborative multimedia in-network processing for data filtering and compression in wireless multimedia sensor networks (WMSNs). This paper proposes an information theoretic image compression framework with an objective to maximize the overall compression of the visual information gathered in a WMSN. The novelty of this framework relies on its independence of specific image types and coding algorithms, thereby providing a generic mechanism for image compression under different coding solutions. The proposed framework consists of two components. First, an entropy-based divergence measure (EDM) scheme is proposed to predict the compression efficiency of performing joint coding on the images collected by spatially correlated cameras. The EDM only takes camera settings as inputs without requiring statistics of real images. Utilizing the predicted results from EDM, a distributed multi-cluster coding protocol (DMCP) is then proposed to construct a compression-oriented coding hierarchy. The DMCP aims to partition the entire network into a set of coding clusters such that the global coding gain is maximized. Moreover, in order to enhance decoding reliability at data sink, the DMCP also guarantees that each sensor camera is covered by at least two different coding clusters. Experiments on H.264 standards show that the proposed EDM can effectively predict the joint coding efficiency from multiple sources. Further simulations demonstrate that the proposed compression framework can reduce 10%–23% total coding rate compared with the individual coding scheme, i.e., each camera sensor compresses its own image independently.

Recently, there has been a great deal of research on using mobility in wireless sensor networks (WSNs) to facilitate surveillance and reconnaissance in a wide deployment area. Besides providing an extended sensing coverage, node mobility along with spatial correlation introduces new network dynamics, which could lead to the traffic patterns fundamentally different from the traditional (Markovian) models. In this paper, a novel traffic modeling scheme for capturing these dynamics is proposed that takes into account the statistical patterns of node mobility and spatial correlation. The contributions made in this paper are twofold. First, it is shown that the joint effects of mobility and spatial correlation can lead to bursty traffic. More specifically, a high mobility variance and small spatial correlation can give rise to pseudo-long-range-dependent (LRD) traffic (high bursty traffic), whose autocorrelation function decays slowly and hyperbolically up to a certain cutoff time lag. Second, due to the ad hoc nature of WSNs, certain relay nodes may have several routes passing through them, necessitating local traffic aggregations. At these relay nodes, our model predicts that the aggregated traffic also exhibits the bursty behavior characterized by a scaled power-law decayed autocovariance function. According to these findings, a novel traffic shaping protocol using movement coordination is proposed to facilitate effective and efficient resource provisioning strategy. Finally, simulation results reveal a close agreement between the traffic pattern predicted by our theoretical model and the simulated transmissions from multiple independent sources, under specific bounds of the observation intervals

The spatial correlation among the images retrieved from distributed video sensors leads to considerable data redundancy, thus resulting in significant performance degradation in energy efficiency and QoS satisfaction. In this paper, a correlation-aware QoS routing algorithm (CAQR) is proposed to efficiently deliver visual information under QoS constraints by exploiting the correlation among video sensors. Firstly, a correlation-aware differential coding scheme is designed to reduce the amount of traffic generated by correlated video sensors. Then, a correlation-aware load balancing scheme is proposed to prevent network congestion by splitting the correlated flows that cannot be reduced to different paths. Finally, these correlation-aware schemes are integrated into an optimization QoS routing framework with an objective to minimize energy consumption subject to QoS constraints. Simulation results show that the proposed algorithm achieves efficient delivery of visual information under QoS constraints in wireless video sensor networks.

Recently, there has been a great deal of research on investigating the effects of mobility on network attributes such as capacity, connectivity, and coverage. In this paper, the node mobility is studied from a new perspective with an objective to reveal the inherent impact of different mobility models on the the traffic patterns in wireless sensor networks. Specifically, the transmission pattern of a mobile sensor node is first characterized by an alternating renewal process that changes states between the active and the inactive. Then, the active state distribution is investigated under four commonly used mobility models: random walk, random waypoint, discrete Brownian motion, and extended Levy walk. For each mobility model, the spectrum of the traffic oriented from a single node is analyzed based on renewal theory. According to this analysis, novel results regarding the impact of each mobility model on the traffic nature are found: random walk, random waypoint, and discrete Brownian motion can only induce short range dependent traffic, whose autocorrelation function decays exponentially fast. In contrast, the traffic under extended Levy walk exhibits pseudo long range dependence, in which the autocorrelation function decays slower than exponential and follows a power law form at large time lags. Finally, the revealed findings are verified by the statistical analysis on the collected traffic traces from the simulated transmissions.

We investigate the dynamic connectivity of cognitive radio ad-hoc networks (secondary networks) coexisting with licensed networks (primary networks) that experience time-varying on-off links. It is shown that there exists a critical density λs*such that if the density of secondary networks is larger than λs*, the secondary network percolates at all time t >; 0, i.e., there exists always an infinite connected component in the secondary network under the time- varying spectrum availability. Furthermore, the upper and lower bounds of λs*are derived and it is shown that they do not depend on the random locations of primary and secondary users, but only on the network parameters, such as active/inactive probability of primary users, transmission range, and the user density.

Data redundancy caused by correlation has motivated the application of collaborative multimedia in-network processing for data filtering and compression in wireless multimedia sensor networks (WMSNs). This paper proposes an information theoretic data compression framework with an objective to maximize the overall compression of the visual information gathered in a WMSN. To achieve this, an entropy-based divergence measure (EDM) scheme is proposed to predict the compression efficiency of performing joint coding on the images collected by spatially correlated cameras. The novelty of EDM relies on its independence of the specific image types and coding algorithms, thereby providing a generic mechanism for prior evaluation of compression under different coding solutions. Utilizing the predicted results from EDM, a distributed multi-cluster coding protocol (DMCP) is proposed to construct a compression-oriented coding hierarchy. The DMCP aims to partition the entire network into a set of coding clusters such that the global coding gain is maximized. Moreover, in order to enhance decoding reliability at data sink, the DMCP also guarantees that each sensor camera is covered by at least two different coding clusters. Experiments on H.264 standards show that the proposed EDM can effectively predict the joint coding efficiency from multiple sources. Further simulations demonstrate that the proposed compression framework can reduce 10% - 23% total coding rate compared with the individual coding scheme, i.e., each camera sensor compresses its own image independently.

Abstract This paper proposes a joint data aggregation and encryption scheme using Slepian‐Wolf coding for efficient and secured data transmission in clustered wireless sensor networks (WSNs). We first consider the optimal intra‐cluster rate allocation problem in using Slepian‐Wolf coding for data aggregation, which aims at finding a rate allocation subject to Slepian‐Wolf theorem such that the total energy consumed by all sensor nodes in a cluster for sending encoded data is minimized. Based on the properties of Slepian‐Wolf coding with optimal intra‐cluster rate allocation, a novel encryption mechanism, called spatially selective encryption, is then proposed for data encryption within a single cluster. This encryption mechanism only requires a cluster head to encrypt its data while allowing all its cluster members to send their data without performing any encryption. In this way, the data from all cluster members can be protected as long as the data of the cluster head (calledvirtual key) is protected. This can significantly reduce the energy consumption for performing data encryption. Furthermore, an energy‐efficient key establishment protocol is also proposed to securely and efficiently establish the key used for encrypting the data of a cluster head. Simulation results show that the joint data aggregation and encryption scheme can significantly improve energy efficiency in data transmission while providing a high level of data security. Copyright © 2009 John Wiley & Sons, Ltd.

Recently there has been a great deal of research on using mobility in wireless sensor networks to facilitate surveillance and reconnaissance in a wide deployment area. Besides providing an extended sensing coverage, the node mobility along with the spatial correlation of the monitored phenomenon introduces new dynamics to the network traffic. These dynamics could lead to long range dependent (LRD) traffic, which necessitates network protocols fundamentally different from what we have employed in the traditional (Markovian) traffic. Therefore, characterizing the effects of mobility and spatial correlation on the dynamic behavior of the network traffic is particularly important in the effective design of network protocols. In this paper, a novel traffic modeling scheme for capturing these dynamics is proposed that takes into account the statistical patterns of human mobility and spatial correlation. The contributions made in this paper are twofold: first, it is shown that the mobility variability and the spatial correlation can lead to the pseudo-LRD traffic, whose autocorrelation function follows a power law form with the Hurst parameter up to a certain cutoff time lag. Second, it is shown that the degree of traffic burstiness, which is characterized by the Hurst parameter, has an intimate connection with the mobility variability and the degree of spatial correlation. Furthermore, we show that this connection can be utilized to design the mobility-aware traffic smoothing schemes, which point out a new direction for traffic control protocols. Finally, simulation results reveal a close agreement between the traffic pattern predicted by our theoretical model and the simulated transmissions from multiple independent sources, under specific bounds of the observation intervals.

Abstract This paper proposes a cooperative fault‐detection mechanism for detecting cluster‐head failures in cluster‐based UnderWater Sensor Networks (UWSNs). The proposed detection mechanism aims to accurately and fast detect the failure of a cluster head in order to avoid unnecessary energy consumption caused by a mistaken detection. For this purpose, it allows each cluster member to independently detect the fault status of its cluster head and then employs a distributed agreement protocol to reach an agreement on the fault status of the cluster head among multiple cluster members. It runs concurrently with normal network operation by periodically performing a detection process at each cluster member. To reduce energy consumption, it uses a time division multiple access medium access control (TDMA MAC) protocol and makes use of the data periodically sent by a cluster head as the heartbeats for fault detection. A couple of forward and backward time‐division‐multiplexing (TDM) frames are specially structured for enabling multiple cluster members to reach an agreement within two frames in each detection process. Moreover, a schedule generation algorithm is also proposed for a cluster head to generate the transmission schedule in the forward and backward frames. Through simulation results, we show that the proposed detection mechanism can achieve high detection accuracy under high packet loss rates in the harsh underwater environment, and can detect a cluster‐head failure faster than a traditional fault‐detection mechanism within a delay bound of two TDM frames. Copyright © 2008 John Wiley & Sons, Ltd.

A Wireless Multimedia Sensor Network (WMSN) is an emerging networking paradigm that allows retrieving video and audio streams, still images, and generic sensing data. Different from conventional wireless sensor networks, a WMSN requires higher network bandwidth and throughput to deliver multimedia contents effectively using energy-constrained devices. In this paper, we propose a clustered on-demand multi-channel MAC protocol (COM-MAQ) to support energy-efficient, high- throughput, and reliable data transmission in WMSNs. The operation of proposed protocol consists of three sessions: request session, scheduling session, and data transmission session. For COM-MAC to achieve high energy efficiency, first, a scheduled multi-channel medium access is used within each cluster so that cluster members can operate in a contention-free manner within both time and frequency domains to avoid collision, idle listening and overhearing. Second, to maximize the network throughput, a traffic-adaptive and QoS-aware scheduling algorithm is executed to dynamically allocate time slots and channels for sensor nodes based on the current data traffic information and QoS requirements. Finally, to enhance transmission reliability, a spectrum-aware ARQ is incorporated to better exploit the unused spectrum for a balance between the reliability and retransmission. Simulation results indicate that COM-MAC can achieve increased network throughput at the cost of a small control and energy overhead.

Node clustering has been widely considered in underwater sensor networks (UWSNs) to improve energy efficiency and prolong network lifetime. Network survivability is a great concern in cluster-based UWSNs. In this paper, we propose a dependable clustering protocol to provide a survivable cluster hierarchy against cluster-head failures in such networks. The proposed clustering protocol attempts to select a primary cluster head and a backup cluster head during clustering so that the cluster members associated with the failed cluster head can quickly switch over to the backup cluster head in the event of a cluster-head failure. Meanwhile, it attempts to select a set of clusters with minimum total cost so that network lifetime can be prolonged to ensure long-term underwater environmental monitoring. Simulation results show that the protocol can effectively enhance network survivability and improve network capacity in the event of cluster-head failures.

In this paper, we propose an efficient fault-prevention clustering protocol for improving the lifetime and robustness of underwater sensor networks (UWSNs). The proposed clustering protocol takes into account both the reliability and residual energy status of each sensor node during clustering, and attempts to select those healthy nodes as cluster heads through failure prediction, cost evaluation, and clustering optimization. The purpose of failure prediction is to predict the potential failure of an underwater sensor based on its lifetime distribution so that those unhealthy nodes are prevented from being selected as cluster heads. Cost evaluation is introduced to evaluate the cost caused by the failure of a cluster head. Clustering optimization aims to construct a cluster hierarchy that minimizes the overall cost of all selected clusters based on the cost evaluation of each sensor node. The simulation results show that the proposed clustering protocol can not only significantly prolong network lifetime, but also improves network robustness and capacity compared with existing clustering protocols.

In this paper, we propose a combined data aggregation and encryption scheme using Slepain-Wolf coding for efficient and secured data transmission in wireless sensor networks (WSNs). We first study the optimal intra-cluster rate allocation problem in using Slepain-Wolf coding for data aggregation, which aims to find a rate allocation subject to Slepian-Wolf theorem such that the total energy consumed by all sensor nodes in the cluster for sending encoded data is minimized. Based on the properties of Slepain-Wolf coding with optimal intra-cluster rate allocation, we then propose a novel encryption mechanism, called spatially selective encryption, for data encryption within a single cluster. This encryption mechanism only requires the cluster head to encrypt its data while allowing all cluster members to send their data without performing any encryption. Using this mechanism, as long as the data of the cluster head (or thevirtualkey) is protected, the data from all cluster members can also be protected, which can significantly reduce the energy consumption for data encryption. Furthermore, an energy-efficient key establishment protocol is also proposed to securely and efficiently establish the key used for encrypting thevisualkey.

In this paper, we study the application of distributed lossy source coding for data aggregation in cluster-based wireless sensor networks (WSNs). We consider a clustered lossy coding (CLC) problem, which aims to select a set of disjoint clusters to cover the whole network such that the total rate of encoded data generated by all clusters or nodes in the network is minimized, given the spatial correlation structure of the network and a couple of total and individual distortion constraints. To solve this problem, we first prove that the overall optimization problem can be decoupled into two independent optimization problems: an optimal clustering problem and an optimal distortion allocation problem. The first problem aims at constructing a clustered hierarchy to minimize the global network entropy without considering distortion allocation, while the second problem aims to optimally allocate a distortion to each sensor node under the given distortion constraints without considering node clustering. We then present a distributed optimal-compression clustering protocol to solve the first problem and use Lagrange multipliers to solve the second problem.

In this paper, we propose an agreement-based fault detection mechanism for detecting cluster-head failures in clustered UnderWater Sensor Networks (UWSNs). The proposed detection mechanism aims to accurately detect the failure of a cluster head in order to avoid unnecessary energy consumption caused by a mistaken detection. For this purpose, it allows each cluster member to independently detect the fault status of its cluster head and at the same time employs a distributed agreement protocol to reach an agreement on the fault status of the cluster head among multiple cluster members. The detection mechanism is based a TDMA MAC protocol used in the network and runs concurrently with normal network operation by periodically performing a distributed detection process at each cluster member. To reduce energy consumption, it makes use of the data periodically sent by a cluster head as the heartbeats for fault detection. A couple of forward and backward TDM frames are specially structured for enabling multiple cluster members to reach an agreement within two frames in each detection process. Moreover, a schedule generation algorithm is also proposed for a cluster head to generate the transmission schedule of the forward and backward frames. Through simulation results, we show that the proposed detection mechanism can achieve high detection accuracy under high packet loss rates in the harsh underwater environment, and can faster detect a cluster-head failure than a traditional fault detection mechanism.

In this paper, we study the node clustering problem in underwater sensor networks (UWSNs). We formulate the problem into a cluster-centric cost-based optimization problem with an objective to improve the energy efficiency and prolong the lifetime of the network. For this purpose, a cost metric is defined for a potential cluster, which takes into account three important parameters that are relevant to the energy status of the cluster, including (1) the total energy consumption of the cluster members for sending data to the cluster head; (2) the residual energy of the cluster head and its cluster members; and (3) the relative location between the cluster head and the underwater sink (uw-sink). To solve the formulated problem, a novel distributed clustering protocol called minimum-cost clustering protocol (MCCP) is proposed, which selects a set of non-overlapping clusters from all potential clusters based on the cost metric assigned to each potential cluster and attempts to minimize the overall cost of the selected clusters. MCCP can adapt geographical cluster head distribution to the traffic pattern in the network and thus avoid the formation of hot spots around the uw-sink. It can also balance the traffic load between cluster heads and cluster members through periodical re-clustering the sensor nodes in the network. Simulation results show that MCCP significantly improves the energy efficiency and the lifetime of a UWSN as compared with the well-known HEED protocol.

Slepian-Wolf coding is a promising distributed source coding technique that can completely remove the data redundancy caused by the spatially correlated observations in wireless sensor networks (WSNs). In this paper, we study the major problems in applying Slepian-Wolf coding for data aggregation in cluster-based WSNs with an objective to optimize data compression so that the total amount of data in the whole network is minimized. We first consider the clustered Slepian-Wolf coding problem, which aims at selecting a set of disjoint potential clusters to cover the whole network such that the global compression gain of Slepian-Wolf coding is maximized. To solve this problem, a distributed optimal-compression clustering protocol (DOC2) is proposed. Under the optimal cluster hierarchy constructed by DOC2, we then consider the optimal intra-cluster rate allocation problem and present an approximation algorithm that can find an optimal rate allocation within each cluster to minimize the intra-cluster communication cost. With the optimal intra-cluster rate allocation found, the procedures to perform Slepian-Wolf coding within a cluster are also presented.