When visual sensor networks (VSNs) enter daily life in society, they not only bring great convenience but also cause people to worry about privacy. Traditional image encryption uses the snowflake effect to protect the privacy, but this can compromise the usability of directly browsing images captured by nodes in VSNs. Recently, thumbnail-preserving encryption (TPE) has been proposed to balance image usability and privacy. However, the Markov chain’s limited connectivity and high time cost, which result in low efficiency, may preclude its application in VSNs. Motivated by this, we propose a novel TPE scheme to protect the image collected in VSNs’ privacy efficiently. We first deeply study the bijective relationship between the three-pixel group and its rank, and then propose a new number method for both, based on which we design the efficient rank mapping. Subsequently, an efficient thumbnail-preserving image encryption via three-pixel rank mapping (E-TPE) is designed, achieving NR security and enhancing the Markov chain’s connectivity. The experiments demonstrate the scheme’s efficiency; that is, it is currently the quickest ideal-TPE scheme with NR security, and a good balance is achieved between usability and privacy.

In the past decades, explosive numbers of Internet of Things (IoT) devices (objects) have been connected to the Internet, which enable users to access, control, and monitor their surrounding phenomenons at anytime and anywhere. To provide seamless interactions between the cyber world and the real world, Digital twins (DTs) of objects (IoT devices) are key enablers for real time monitoring, behaviour simulations, and predictive decisions on objects. Compared to centralized cloud computing, mobile edge computing (MEC) has been envisioning as a promising paradigm for low latency IoT applications. Accelerating the usage of DTs in MEC networks will bring unprecedented benefits to diverse services, through the co-evolution between physical objects and their virtual DTs, and DT-assisted service provisioning has attracted increasing attention recently. In this article, we consider novel DT placement and migration problems in an MEC network with the mobility assumption of objects and users, by jointly considering the freshness of DT data and the service cost of users requesting for DT data. To this end, we first propose an algorithm for the DT placement problem with the aim to minimize the sum of the DT update cost of objects and the total service cost of users requesting for DT data, through efficient DT placements and resource allocation to process user requests. We then devise an approximation algorithm with a provable approximation ratio for a special case of the DT placement problem when each user requests the DT data of only one object. Meanwhile, considering the mobility of users and objects, we devise an online, two-layer scheduling algorithm for DT migrations to further reduce the total service cost of users within a given finite time horizon. We finally evaluate the performance of the proposed algorithms through experimental simulations. The simulation results show that the proposed algorithms are promising.

UWB (Ultra-wideband) has been shown to be a promising technology to provide accurate positioning for the Internet of Things. However, its performance significantly degrades in practice due to Non-Line-Of-Sight (NLOS) issues. Various approaches have implicitly or explicitly explored the problem. In this article, we propose RefLoc , which leverages the unique benefits of UWB to address the NLOS problem. While we find that NLOS links can vary significantly in the same environment, LOS links possess similar features that can be captured by the high bandwidth of UWB. Specifically, the high-level idea of RefLoc is to first identify links among anchors with known positions and leverage those links as references for tag link identification. To achieve this, we address the practical challenges of deriving anchor link status, extracting qualified link features, and inferring tag links with anchor links. We implement RefLoc on commercial hardware and conduct extensive experiments in different environments. The evaluation results show that RefLoc achieves an average NLOS identification accuracy of 96% in various environments, improving the state-of-the-art by 10%, and reduces 80% localization error with little overhead.

The use of drones for collecting information and detecting bugs in orchards covered by nets is a challenging problem. The nets help in reducing pest damage, but they also constrain the drone’s flight path, making it longer and more complex. To address this issue, we model the orchard as an aisle-graph, a regular data structure that represents consecutive aisles where trees are arranged in straight lines. The drone flies close to the trees and takes pictures at specific positions for monitoring the presence of bugs, but its energy is limited, so it can only visit a subset of positions. To tackle this challenge, we introduce the Single-drone Orienteering Aisle-graph Problem (SOAP), a variant of the orienteering problem, where likely infested locations are prioritized by assigning them a larger profit. Additionally, the drone’s movements have a cost in terms of energy, and the objective is to plan a drone’s route in the most profitable locations under a given drone’s battery. We show that SOAP can be optimally solved in polynomial time, but for larger orchards/instances, we propose faster approximation and heuristic algorithms. Finally, we evaluate the algorithms on synthetic and real datasets to demonstrate their effectiveness and efficiency.

Mobile crowdsensing leverages the power of a vast group of participants to collect sensory data, thus presenting an economical solution for data collection. However, due to the variability among participants, the quality of sensory data varies significantly, making it crucial to extract truthful information from sensory data of differing quality. Additionally, given the fixed time and monetary costs for the participants, they typically only perform a subset of tasks. As a result, the datasets collected in real-world scenarios are usually sparse. Current truth discovery methods struggle to adapt to datasets with varying sparsity, especially when dealing with sparse datasets. In this article, we propose an adaptive Hypergraph-based EM truth discovery method, HGEM. The HGEM algorithm leverages the topological characteristics of hypergraphs to model sparse datasets, thereby improving its performance in evaluating the reliability of participants and the true value of the event to be observed. Experiments based on simulated and real-world scenarios demonstrate that HGEM consistently achieves higher predictive accuracy.

This study introduces an innovative method aimed at reducing energy consumption in sensor networks by predicting sensor data, thereby extending the network’s operational lifespan. Our model, Temporal Graph Sensor Prediction (TG-SPRED), predicts readings for a subset of sensors designated to enter sleep mode in each time slot, based on a non-scheduling-dependent approach. This flexibility allows for extended sensor inactivity periods without compromising data accuracy. TG-SPRED addresses the complexities of event-based sensing—a domain that has been somewhat overlooked in existing literature—by recognizing and leveraging the inherent temporal and spatial correlations among events. It combines the strengths of Gated Recurrent Units and Graph Convolutional Networks to analyze temporal data and spatial relationships within the sensor network graph, where connections are defined by sensor proximities. An adversarial training mechanism, featuring a critic network employing the Wasserstein distance for performance measurement, further refines the predictive accuracy. Comparative analysis against six leading solutions using four critical metrics—F-score, energy consumption, network lifetime, and computational efficiency—showcases our approach’s superior performance in both accuracy and energy efficiency.

Single-modal data has a limitation on fatigue detection, while the shortage of labeled data is pervasive in multimodal sensing data. Besides, it is a time-consuming task for board-certified experts to manually annotate the physiological signals, especially hard for EEG sensor data. To solve this problem, we propose FedUSL (Federated Unified Space Learning), a federated annotation method for multimodal sensing data in the driving fatigue detection scenario, which has the innate ability to exploit more than four multimodal data simultaneously for correlations and complementary with low complexity. To validate the efficiency of the proposed method, we first collect the multimodal data (aka, camera, physiological sensor) through simulated fatigue driving. The data is then preprocessed and features are extracted to form a usable multimodal dataset. Based on the dataset, we analyze the performance of the proposed method. The experimental results demonstrate that FedUSL outperforms other approaches for driver fatigue detection with carefully selected modal combinations, especially when a modality contains only \(10\% \) labeled data.

With the help of 5G network, edge intelligence (EI) can not only provide distributed, low-latency, and high-reliable intelligent services, but also enable intelligent maintenance and management of smart city. However, the constantly changing available computing resources of end devices and edge servers cannot continuously guarantee the performance of intelligent inference. In order to guarantee the sustainability of intelligent services in smart city, we propose the Adaptive Model Selection and Partition Mechanism (AMSPM) in 5G smart city where EI provides services, which mainly consists of Adaptive Model Selection (AMS) and Adaptive Model Partition (AMP). In AMSPM, the model selection and partition of deep neural network (DNN) are formulated as an optimization problem. Firstly, we propose a recursive-based algorithm named AMS based on the computing resources of edge devices to derive an appropriate DNN model that satisfies the latency demand of intelligent services. Then, we adaptively partition the selected DNN model according to the computing resources of edge devices. The experimental results demonstrate that, when compared with state-of-the-art model selection and partition mechanisms, AMSPM not only reduces latency but also enhances computing resource utilization.

Small Unmanned Aerial Vehicles (UAVs) are becoming potential threats to security-sensitive areas and personal privacy. A UAV can shoot photos at height, but how to detect such an uninvited intruder is an open problem. This article presents mmHawkeye, a passive approach for non-cooperative UAV detection and identification with a commercial off-the-shelf millimeter wave (mmWave) radar. mmHawkeye does not require prior knowledge of the type, motions, and flight trajectory of the UAV, while exploiting the signal feature induced by the UAV’s periodic micro-motion (PMM) for long-range accurate detection. The design is therefore effective in dealing with low signal-to-noise ratio and uncertain reflected signals from the UAV. After analyzing the theoretical model of the PMM feature, mmHawkeye can further track the UAV’s position containing range, azimuth and altitude angle with dynamic programming and particle filtering and then identify it with a Long Short-Term Memory–based detector. We implement mmHawkeye on a commercial mmWave radar and evaluate its performance under varied settings. The experimental results show that mmHawkeye has a detection accuracy of 95.8% and can realize detection at a range up to 80 m.

As an emerging mobility-on-demand service, bike-sharing system (BSS) has spread all over the world by providing a flexible, cost-efficient, and environment-friendly transportation mode for citizens. Demand-supply unbalance is one of the main challenges in BSS because of the inefficiency of the existing bike repositioning strategy, which reallocates bikes according to a pre-defined periodic schedule without considering the highly dynamic user demands. While reinforcement learning has been used in some repositioning problems for mitigating demand-supply unbalance, there are significant barriers when extending it to BSS due to the dimension curse of action space resulting from the dynamic number of workers and bikes in the city. In this paper, we study these barriers and address them by proposing a novel bike repositioning system, namely BikeBrain, which consists of a demand prediction model and a spatio-temporal bike repositioning algorithm. Specifically, to obtain accurate and real-time usage demand for efficient bike repositioning, we first present a prediction model ST-NetPre, which directly predicts user demand considering the highly dynamic spatio-temporal characteristics. Furthermore, we propose a spatio-temporal cooperative multi-agent reinforcement learning method (ST-CBR) for learning the worker-based bike repositioning strategy in which each worker in BSS is considered an agent. Especially, ST-CBR adopts the centralized learning and decentralized execution way to achieve effective cooperation among large-scale dynamic agents based on Mean Field Reinforcement Learning (MFRL), while avoiding the huge dimension of action space. For dynamic action space, ST-CBR utilizes a SoftMax selector to select the specific action. Meanwhile, for the benefits and costs of agents’ operation, an efficient reward function is designed to seek an optimal control policy considering both immediate and future rewards. Extensive experiments are conducted based on large-scale real-world datasets, and the results have shown significant improvements of our proposed method over several state-of-the-art baselines on the demand-supply gap and operation cost measures.