Radio Frequency Identification (RFID) technology plays a pivotal role in modern applications ranging from retail and logistics to healthcare and security. However, a fundamental challenge persists in large-scale RFID systems: maximizing the coverage capacity of readers – the ability to reliably identify and communicate with the maximum number of tags within their operational range. While previous research has explored various aspects of RFID performance, the systematic optimization of coverage capacity remains underinvestigated. This paper addresses this gap by developing a comprehensive framework that integrates theoretical analysis with practical implementation strategies. We first establish a novel coverage capacity model that incorporates critical factors such as tag spatial distribution and link loss dynamics, providing a theoretical foundation for determining the upper bounds of reader performance. Building on this, we propose a cutoff-power-based optimization approach that dynamically adapts to real-world conditions without relying on predefined system parameters. Furthermore, to extend coverage across larger areas, we investigate advanced multi-reader configurations and present strategic deployment methodologies. Our framework is designed to characterize the baseline coverage capacity of existing RFID readers, rather than to enlarge the interrogation zone through additional hardware or protocol modifications. Moreover, it can be naturally extended to advanced configurations such as MIMO or phased-array systems once their antenna parameters are specified. The effectiveness of our approach is validated through extensive experiments using commercial RFID hardware, demonstrating measurable improvements in coverage capability. By bridging theoretical principles with practical constraints, this work offers actionable insights for designing and deploying high-performance RFID systems.

Recent works in domain adaptive object detection (DAOD) have demonstrated integrating domain adaptation modules into detectors can improve the models' transferability. Nevertheless, given the presence of multiple objects within an image and uncertainties in their localizations, adapting features to diverse objects faces a potential challenge in distinguishing foreground from background. This may hurt the detector's discriminability. Furthermore, traditional methods solely rely on category information for foreground object identification, thereby overlooking the significant category-agnostic information, namely objectness features, and leading to a negative transfer. In this paper, we propose a novel collaborative architecture named Progressively Decoupled adaptation with Objectness Information (PDOI), which individually adapts proposals and semantics, and decouples them from the training of the detector. PDOI sequentially cascades three parts: a category-agnostic adaptor, a category-conditional adaptor, and a detector. The first adaptor focuses on learning category-agnostic objectness features to enhance cross-domain proposal alignments. Subsequently, the category-conditional adaptor is designed to achieve cross-domain semantic alignments by adaptively learning the semantic attributes within objectness features. Finally, we train the detector using pseudo-proposal/category labels generated by two adaptors. We develop a self-feedback optimization mechanism enabling mutual enhancement between the two adapters and detectors. Our approach consistently outperforms state-of-the-art methods across several DAOD benchmarks.

Due to the adaptable deployment and extensive coverage, autonomous aerial vehicles (AAVs) have gained considerable attention in the field of mobile edge computing (MEC). Nevertheless, meeting the requirements of task offloading in AAV-assisted MEC remains challenging due to limitations in computing power and battery life. To address this challenge, we investigate the problem of distributed joint task partial offloading and energy harvesting (EH) in AAV-assisted MEC to minimize both task execution latency and energy cost of the AAV system. We formulate this problem as a mixed-integer nonlinear programming problem and decompose this problem into two subproblems: 1) task offloading strategy problem and 2) task offloading ratio and EH problem. We propose a distributed algorithm to determine task offloading strategies of AAVs iteratively. Given the task offloading strategies of AAVs, the subproblem of task offloading ratio and EH duration is solved by the sine cosine algorithm. Experimental results demonstrate that our proposed algorithm can effectively reduce the task execution latency and energy cost of the AAV-assisted MEC system, and outperforms other baselines significantly.

Decentralized Applications (DApps) have garnered significant attention due to their decentralization, anonymity, and data autonomy. However, these systems face potential privacy challenge. The privacy challenge arises from the necessity for external service providers to collect and process user interaction data. The untrustworthiness of these providers may lead to privacy breaches, compromising the overall security of such DApp environments. To address this challenge, we model the interaction data in the DApp environments as dynamic graphs and propose a dynamic graph publication method named HMG (Hidden Markov Model for Dynamic Graphs). HMG estimates the interaction probabilities between users by extracting the temporal information from historically collected data and constructs an optimized model to generate synthetic graphs. The synthetic graphs can preserve the dynamic topological characteristics of the interaction processes within DApp environments while effectively protecting user privacy, thus assisting external service providers in performing effective analyses. Finally, we evaluate the performance of HMG using real-world datasets and benchmark it against commonly used graph metrics. The results demonstrate that the synthetic graphs preserve essential features, making them suitable for analysis by service providers.

Serverless computing is rapidly evolving as a new cloud computing paradigm, where organizing applications into function chains is a key characteristic that enhances the adaptability of serverless frameworks. However, serverless platforms experience performance degradation due to the cascading cold starts of function chains. Current research mainly addresses cold starts by reducing their duration and frequency, but often neglects the ubiquitous dependencies among functions. In this paper, we introduce Thawbringer, a serverless function chain orchestrator. Thawbringer first employs a multi-layer Bidirectional Gated Recurrent Unit (BiGRU) algorithm to extract invocation pattern features from function invocation history data abundant in temporal information. It then combines the Pearson correlation coefficient with agglomerative hierarchical clustering to identify dependency profiles from the function invocation data. Utilizing these profiles, Thawbringer generates function chains and performs fine-grained scheduling to minimize the occurrence of cascading cold starts. Our empirical system evaluation, using representative serverless applications and industry-grade workload traces, demonstrates that Thawbringer reduces function cold start rates by 31.67%, container memory wastage by 25.65%, and end-to-end latency by 26.79% compared to state-of-the-art solutions.

Serverless introduces a lightweight, function-based execution model that is significant in addressing challenges such as heterogeneity in Internet of Things (IoT) edge applications, high dynamics in user requests, and unpredictability in workloads. Firstly, in this paper, we propose a novel scaling algorithm evaluation metric, the economic model Return on Investment (ROI). This metric encompasses elements such as response latency, latency stability, function invocation, and resource usage to assess the Quality of Service (QoS) per unit of monetary cost for application providers. Secondly, we introduce RIA, a serverless edge framework that serves as a high ROI auto-scaler for edge functions. RIA incorporates the SHAP-based DQN (SDQN) algorithm, integrating threshold-based reactive methods with prediction-based proactive approaches to make optimal scaling decisions based on the current state of the edge environment. It includes a SHAP-based space optimization algorithm, effectively addressing the issue of state space explosion caused by high volatility and numerous concurrent functions in traditional reinforcement learning in edge environments. Finally, we conduct extensive experiments based on Azure traces to evaluate the effectiveness and performance of RIA. We compare RIA with five state-of-the-art technologies. The experimental results demonstrate a reduction in QoS violations by 25.66%-83.08%, an increase in ROI by 0.86 to 6.45 times, and the second-lowest monetary cost.

Mobile edge computing introduces a novel computing paradigm for mobile devices, reducing execution latency and energy consumption by offloading tasks to edge servers or other idle mobile devices. In this paper, we consider the utility optimization problem of two typical computing tasks, latency-sensitive tasks and latency-tolerant tasks, among multiple mobile devices and base stations. Mobile devices can choose three computing modes to optimize utility: local computing, task allocation to base stations, and task allocation to other mobile devices through device-to-device communication. To address this problem, we formalize it as a potential game for multi-mobile device multi-base station task offloading. Furthermore, we prove the existence of a Nash equilibrium for the modeled potential game and propose a task allocation scheme for hybrid tasks. This scheme maximizes both energy consumption utility and task execution utility by optimizing task offloading mode selection and task execution order scheduling. Simulation results show that our proposed scheme can substantially enhance user utility and has good scalability with the increase of mobile devices.

The deployment and management of distributed latency-sensitive applications pose significant challenges due to the volatile nature of modern Internet of Things (IoT) edge networks, constrained resources for heterogeneous edge devices, and the increasingly demanding user requirements for service level agreement (SLA) and quality of service (QoS). To address these challenges, Kubernetes (k8s) orchestration platform has emerged, with previous research using heuristics to make scheduling decisions quickly or leveraging artificial intelligence (AI) to adapt to changing circumstances. However, the former often struggle to adapt to highly dynamic and resource-competitive edge environments, while the latter’s long decision latency negatively impacts QoS. In this paper, we propose NeiLatS, a heterogeneous cloud-edge cluster scheduling system that addresses these issues. NeiLatS can monitor real-time resource and communication link conditions and make the scheduler latency-aware and location-aware. Additionally, we innovatively propose an efficacy coefficient method with tolerance factor ε (ε -ECM), which fully accounts for multi-resource load balancing, network stability, and future communication link conditions to resolve the conflict between resource utilization balance and QoS. We test NeiLatS on physical clusters and compare it with five different algorithms. Our evaluation results show a 4.59%-16% improvement in resource load balancing degree, a 41.95%-82.85% reduction in SLA violation rates, and a 1-5x improvement in QoS, which greatly demonstrate the effectiveness of our proposed method.

Distracted driving behavior is known as a leading factor in road traffic injuries and deaths. Fortunately, rapidly developing deep learning technology has shown its potential in distracted driving detection. Nevertheless, deep learning-based solutions need to collect large amounts of driving data captured by camera sensors in the vehicle, which will cause serious privacy concerns. As a privacy-preserving distributed learning paradigm, federated learning (FL) has achieved competitive performance in many applications recently. Inspired by this, we introduce FL into distracted driving detection tasks. However, we observe that the heterogeneous data distribution across drivers leads to significant performance degradation of the model learned in FL. To address this challenge, we propose a simple and effective federated bidirectional knowledge distillation framework, FedBiKD. Specifically, FedBiKD utilizes the knowledge from the global model in guiding local training to mitigate the issue of local deviation. Meanwhile, the consensus from the ensemble of local models is also employed to fine-tune the aggregated global model for less volatility in training. Our extensive experiments demonstrate the effectiveness of FedBiKD in distracted driving detection. The results show that FedBiKD significantly outperforms other FL algorithms in terms of accuracy, communication efficiency, convergence rate, and stability.

In the real world, speaker recognition systems usually suffer from serious performance degradation due to the domain mismatch between training and test conditions. To alleviate the harmful effect of domain shift, unsupervised domain adaptation methods are introduced to learn domain-invariant speaker representations, which focus on addressing the single-source-to-single-target domain adaptation issue. However, labeled speaker data are usually collected from multiple sources, such as different languages, genres and devices. The single-domain adaptation methods can not deal with the complex multiple-domain mismatch problem. To address this issue, we propose a multiple-domain adaptation framework named CentriForce to extract domain-invariant speaker representations for speaker recognition. Different from previous methods, CentriForce learns multiple domain-related speaker representation spaces. To mitigate the multiple-domain mismatch, CentriForce reduces the Wasserstein distance between each pair of source and target domains in their domain-related representation space and meanwhile uses the target domain as an anchor point to draw all source domains closer to each other. In our experiments, CentriForce achieves the best performance on most of the 16 challenging adaptation tasks, compared with other competing adaptation methods. Ablation study and representation visualization further demonstrate its effectiveness for learning the domain-invariant speaker embedding.

Recent breakthroughs in deep neural networks (DNNs) have fueled a tremendously growing demand for bringing DNN-powered intelligence into mobile platforms. While the potential of deploying DNNs on resource-constrained platforms has been demonstrated by DNN compression techniques, the current practice suffers from two limitations: 1) merely stand-alone compression schemes are investigated even though each compression technique only suit for certain types of DNN layers; and 2) mostly compression techniques are optimized for DNNs’ inference accuracy, without explicitly considering other application-driven system performance (e.g., latency and energy cost) and the varying resource availability across platforms (e.g., storage and processing capability). To this end, we propose AdaDeep, a usage-driven, automated DNN compression framework for systematically exploring the desired trade-off between performance and resource constraints, from a holistic system level. Specifically, in a layer-wise manner, AdaDeep automatically selects the most suitable combination of compression techniques and the corresponding compression hyperparameters for a given DNN. Thorough evaluations on six datasets and across twelve devices demonstrate that${\sf AdaDeep}$can achieve up to$18.6\times$latency reduction,$9.8\times$energy-efficiency improvement, and$37.3\times$storage reduction in DNNs while incurring negligible accuracy loss. Furthermore,${\sf AdaDeep}$also uncovers multiple novel combinations of compression techniques.

In this paper, we focus on designing and developing ProMETheus, an intelligent system for meeting minutes generated from audio data. The first task in ProMETheus is to recognize the speakers from noisy audio data. Speaker recognition algorithm is used to automatically identify who is speaking according to the speech in an audio data. Naturally, speech recognition will transcribe speakers' audio to text so that ProMETheus can generate the complete meeting text with speakers' name chronologically. In order to show the subject of the meeting and the agreed action, we use text summarization algorithm that can extract meaningful key phrases and summary sentences from the complete meeting text. In addition, sentiment analysis for meeting text of different speakers can make the agreed action more humane due to calculating the relevance score of each course by the sentiment and attitude in text tone. The ProMETheus is capable of accurately summarizing the meeting and analyzing the agreed action. Our robust system is evaluated on a real-world audio meeting dataset that involves multiple speakers in each meeting session.

Recent research has demonstrated the potential of deploying deep neural networks (DNNs) on resource-constrained mobile platforms by trimming down the network complexity using different compression techniques. The current practice only investigate stand-alone compression schemes even though each compression technique may be well suited only for certain types of DNN layers. Also, these compression techniques are optimized merely for the inference accuracy of DNNs, without explicitly considering other application-driven system performance (e.g. latency and energy cost) and the varying resource availabilities across platforms (e.g. storage and processing capability). In this paper, we explore the desirable tradeoff between performance and resource constraints by user-specified needs, from a holistic system-level viewpoint. Specifically, we develop a usage-driven selection framework, referred to as AdaDeep, to automatically select a combination of compression techniques for a given DNN, that will lead to an optimal balance between user-specified performance goals and resource constraints. With an extensive evaluation on five public datasets and across twelve mobile devices, experimental results show that AdaDeep enables up to 9.8x latency reduction, 4.3x energy efficiency improvement, and 38x storage reduction in DNNs while incurring negligible accuracy loss. AdaDeep also uncovers multiple effective combinations of compression techniques unexplored in existing literature.

The information gradient-based routing protocols have been proved to be economical and effective by adopting the principle of achieving the global objective through local decision, but lightweight methods to construct the information gradient should be fully investigated, especially in a large-scale network with high information dynamics. In this paper, we focus on the construction of the information gradient by balancing convergence conditions and energy consumption. Therefore, two algorithms, Hierarchical Skeleton-based Construction Algorithm (HSCA) and Estimate value Substitution Algorithm (ESA) are proposed to achieve the goal of fastening the convergence in an energy efficient way. Both of the algorithms obey the typical assumptions on WSNs settings and the gossip-styled propagation principle. Comprehensive simulation results show that the proposed algorithms can reduce iteration times to reach a convergence status by 80% and conserve 30–50% energy consumption on average.

An activity-diary system, named Healthy, is presented in this paper. Healthy can infer users diary of physical activities and energy expenditure based on METS (Metabolic Equivalents) values via recognizing general human activities. In this system, we design a two-layer classifier which costs less energy and memory with satisfactory accuracy. Our classifier divides the activities into two categories: periodic and nonperiodic. And a different sub-classifier is applied for each category. Meanwhile, We design a state listener to recognize more complicated activities. To further improve recognition accuracy, in the second layer sub-classifier, we put forward an adaptive framing algorithm based on the period length of periodical activities to determine the time during which features are extracted. By testing Healthy in real situation, we obtained an average recognition accuracy of 98.0%.

We study the rendezvous data collection problem for the mobile sink in wireless sensor networks. We introduce to jointly optimize trajectory planning for the mobile sink and workload balancing for the network. By doing so, the mobile sink is able to efficiently collect network-wide data within a given delay bound and the network can eliminate the energy bottleneck to dramatically prolong its lifetime. Such a joint optimization problem is shown to be NP-hard and we propose an approximation algorithm, named RPS-LB, to approach the optimal solution. In RPS-LB, according to observed properties of the median reference structure in the network, a series of Rendezvous Points (RPs) are selected to construct the trajectory for the mobile sink and the derived approximation ratio of RPSLB guarantees that the formed trajectory is comparable with the optimal solution. The workload allocated to each RP is proven to be balanced mathematically. We then relax the assumption that mobile sink knows the location of each sensor node and present a localized, fully distributed version, RPS-LB-D, which largely improves the system applicability in practice. We verify the effectiveness of our proposals via extensive experiments.

Target tracking plays a critical role in the applications of wireless sensor networks. In this paper, we propose a target tracking algorithm based on Bayesian estimation, the key steps involved in the algorithm include target detection based on a probabilistic model, preliminary localization using the two-stage target locating algorithm, further prediction adopting Bayesian estimation. Simulation results show that our simple and effective algorithm can accurately track the target and save much energy during the tracking process of a moving target.

For some sensor network applications, the problem of sweep coverage, which periodically covers POIs (Points of Interest) to sense events, is of importance. How to schedule minimum number of mobile sensors to achieve the sweep coverage within specified sweep period is a challenging problem, especially when the POIs to be scanned exceeds certain scale and the speed of mobile sensor is limited. Therefore, multiple mobile sensors are required to collaboratively complete the scanning task. When the mobile sensor is restricted to follow the same trajectory in different sweep periods, we design a centralized algorithm, MinExpand, to schedule the scan path. When the scan path of the existing mobile sensors has been exceeds the length constraint, MinExpand gradually deploys more mobile sensors and eventually achieves sweep coverage to all POIs. When the mobile sensors are not restricted to follow the same trajectory in different sweep periods, we design OSweep algorithm, where all the mobile sensors are scheduled to move along a TSP (Traveling Salesman Problem) ring consists of POIs. We conduct comprehensive simulations to study the performance of the proposed algorithms. The simulation results show that MinExpand and OSweep outperform CSWEEP in both effectiveness and efficiency.

In order to maximize the lifetime of wireless sensor networks, while ensuring the monitoring quality for specific applications, we develop the sleep-wakeup algorithms for the sensor networks. Firstly, for the three-dimensional wireless sensor networks, we formulate the virtual barrier coverage problems. Secondly, the design and implementation of the nodes sleep-wakeup scheduling algorithm, FWP, to maximize the covering time of a single virtual barrier and the k-virtual barriers sleep-wakeup scheduling algorithm, OBP, to maximize the network coverage time of k virtual barriers are presented. Finally, through the comprehensive simulation, the effectiveness of the algorithms is validated. The relationships among the density of the virtual barriers, virtual lattice point, the number of sensors and the sensing radius are also investigated and simulated.

We present Score, a sensor core framework for cross-layer design in wireless sensor networks. Network components running in the context of Score have the ability to collaborate without the need for pair-wise interfaces. This collaboration promotes protocol optimization in the resource constrained wireless sensor networks, a technique widely known as cross-layer design. We also demonstrate the advantage of Score through three example network components.