IET Wireless Sensor Systems最新文献

Efficient resource management remains a critical challenge in wireless sensor networks (WSNs) due to the constrained nature of sensor nodes. This paper proposes a novel hybrid clustering protocol to address this issue, aiming to optimise energy consumption, extend network lifetime and enhance scalability. Our approach combines the improved version of binary dragonfly algorithm (IVBDA) for cluster head (CH) selection and the Mamdani fuzzy inference system for effective cluster formation. After CH selection and cluster formation, a multi-hop routing mechanism transmits data packets within the WSN. To validate the performance of the proposed protocol, extensive simulations are conducted on various network topologies, evaluating metrics such as average energy consumption, live node count, network lifetime, and packet reception at the base station (BS). Comparative analyses with existing clustering protocols and other metaheuristic algorithms, including binary particle swarm optimisation (BPSO), binary whale optimisation algorithm (BWOA) and binary dragonfly algorithm (BDA), demonstrate the superior performance of the proposed hybrid approach in terms of energy efficiency, network longevity and overall WSN performance. The improved version of BDA shows faster convergence than BPSO, BWOA and BDA, as ascertained by examining the multi-objective fitness function. This paper contributes significantly to the development of efficient clustering protocols and showcases the potential of hybrid metaheuristic and fuzzy inference techniques for optimising resource allocation in WSNs. The proposed protocol outperforms other protocols in network lifetime and overall performance, indicating its potential to be a valuable solution for resource management in WSNs. The evaluation of metaheuristic algorithms highlights the importance of considering convergence speed in optimising energy-efficient clustering.

Vital signs are crucial indicators of an individual's physiological well-being and represent one of the primary evaluations conducted in clinical and hospital environments. A comprehensive evaluation of a patient's health state depends on these signs which include heart rate (HR), respiratory rate (RR), blood oxygen saturation (SpO2), blood pressure (BP) and body temperature (BT). In recent years, there has been significant interest in using imaging photoplethysmography (iPPG) with consumer-level cameras for contactless health monitoring (CHM) to accurately assess vital signs. The introduction of iPPG in CHM signifies the beginning of a remarkable era in the history of healthcare, whereby diagnostic processes are enhanced via the integration of technology and patient well-being. This review article presents a comprehensive examination of CHM techniques utilising machine learning (ML) and deep learning (DL) algorithms for the assessment of critical vital signs. The article addresses the challenges and research gaps identified in recent studies, particularly those related to variations in lighting conditions, head movements and the impact of different colour types on the accuracy and reliability of CHM techniques. Finally, we propose several recommendations aimed to enhance the efficiency of CHM systems. These include the development of more robust learning algorithms and the creation of diverse datasets that encompass a wide range of demographics including variations in gender, skin colour and lighting conditions.

The demand for fast, precise, and sensitive food safety methods is growing as consumers increasingly rely on online food delivery services. Food products shipped from South America to Europe can spend more than 21 days in transit, often resulting in deterioration, mould, or pathogen growth. Similar risks apply in the food service industry, where food may have been stored or handled improperly, leading to foodborne illness. This paper presents the Edispotter, an IoT-based food quality monitoring system designed to address these and similar issues. Using Raspberry Pi and ESP32, the Edispotter collects essential food quality data through various sensors. The data are processed and stored in a Redis database within an Amazon Web Services (AWS) cloud environment, providing real-time food-based status updates via an Android application.

The Gaussian mixture probability hypothesis density (GM-PHD) filter is an almost exact closed-form approximation to the Bayes-optimal multi-target tracking algorithm. Due to its optimality guarantees and ease of implementation, it has been studied extensively in the literature. However, the challenges involved in implementing the GM-PHD filter efficiently in a distributed (multi-sensor) setting have received little attention. The existing solutions for distributed PHD filtering either have a high computational and communication cost, making them infeasible for wireless sensor networks with limited communication bandwidths, and/or are unable to guarantee the asymptotic convergence of the algorithm to an optimal solution. In this paper, we develop a distributed GM-PHD filtering recursion that uses a probabilistic communication rule to limit the communication bandwidth of the algorithm, while ensuring asymptotic convergence of the algorithm. The proposed algorithm uses weighted average consensus of Gaussian mixtures (GMs) to lower (and asymptotically minimise) the Cauchy–Schwarz divergences between the sensors' local estimates. In addition, the proposed probabilistic communication rule is able to avoid the issue of false positives, which has previously been noted to impact the filtering performance of distributed multi-target tracking. Through numerical simulations, it is demonstrated that our proposed method is an effective solution for distributed multi-target tracking in resource-constrained sensor networks.

Stress, a common response to challenging situations, has become pervasive in contemporary daily life due to various factors. Persistent stress can weaken the human immune system, increasing the risk of chronic stress and contributing to a range of physical and mental health disorders. Therefore, timely detection of stress in its early stages is crucial for preventing adverse health outcomes. Physiological signals offer insights into the body's stress-induced changes and can be leveraged for stress detection applications. Among these signals, the photoplethysmogram (PPG) signal stands out due to its advantages. This article introduces an innovative stress detection model based on dilated convolutional neural networks (Dilated CNNs), a deep learning algorithm. This model distinguishes between an individual's stressed and non-stressed states by analysing PPG signals without requiring pre-processing, denoising, or feature extraction. Leveraging the Empatica E4 PPG signals from the Wearable Stress and Affect Detection (WESAD) dataset, the authors developed and evaluated the model, achieving remarkable results: a test accuracy of 93.56% and an area under the curve (AUC) of 96.52%. These outcomes are particularly noteworthy given the streamlined data preparation process and methodological simplicity. Beyond enabling early stress diagnosis, this advancement holds promise for enhancing overall health and well-being in the fast-paced and intricate world. Additionally, its simplicity makes it suitable for real-time stress detection and integration into wearable devices.

Quality of service (QoS) and energy efficiency are two major factors that play an important role in wireless sensor network (WSNs) operation. Although it is often argued that these two factors are naturally consistent. WSNs demand additional QoS measures beyond the capabilities of clustering and routing protocols, such as stability and latency. This paper proposes a new routing protocol named improving quality of service of wireless sensor networks using a smart mobile data collector (I-QoS-WSN-S-MDC). I-QoS-WSN-S-MDC is an enhancement of the low energy adaptive clustering hierarchy-kmeans-grid (LEACH-K-G) and the mobile data collector-K-means (MDC-K) to find the optimal path taken by the MDC for QoS efficiency. Specifically, the proposed I-QoS-WSN-S-MDC protocol uses the K-means algorithm and the grid clustering algorithm to reduce energy consumption in the cluster head (CH) election stage. In addition, the MDC is used as an interface between the CH and the base station (BS) to improve the WSN QoS and transmission phase of the MDC-K and LEACH-G-K protocols using lin–kernighan–helsgaun-travelling salesman problem (LKH-TSP). The experimental results show that I-QoS-WSN-S-MDC outperforms several low energy adaptive clustering (LEACH) protocol enhancements such as threshold-sensitive energy efficient network (TEEN), LEACH-K, LEACH-C, Improved-LEACH, Stable-Improved-LEACH, MDC maximum distance leach, MDC minimum distance leach, MDC-K, and MDC-TSP-LEACH-K.

In recent years, indoor localisation based on channel state information (CSI) fingerprint has been actively researched because of the rapid growth of the Internet of Things (IoT). In addition, various deep learning (DL) methods such as deep neural networks (DNN) and convolutional neural networks (CNN) have been widely discussed for the indoor localisation. The CSI-fingerprint can be produced by continuous and quantised values. For the CSI-fingerprint using quantised values, good performance is achieved. However, since quantised data for the optimal level has not been sufficiently discussed, the best performance of quantisation is not indicated. Therefore, in this paper, we propose an effective quantised CSI-fingerprint for DL-based indoor localisation.

This research presents a novel square hollow-core photonic crystal fibre (PCF) sensor designed for the detection of food-grade oils in the terahertz (THz) frequency range. The sensor’s effectiveness is quantitatively evaluated using COMSOL Multiphysics, a sophisticated simulation tool that employs finite element methodology (FEM) to model complex interactions within the fibre structure. Simulation outcomes reveal that, under optimal geometric parameters, the proposed sensor achieves an exceptional relative sensitivity of 98.27% for various edible oils at an ideal frequency of 2.2 THz, significantly outperforming existing technologies. Additionally, the sensor exhibits minimal confinement loss of 1.428 × 10⁻⁸ dB/m and a low effective material loss of 0.004246 cm⁻¹, facilitating accurate detection of slight refractive index variations related to the chemical compositions of different oils. This high sensitivity enables non-destructive testing, allowing for the analysis of oils without compromising their composition or quality, thereby maintaining the integrity of food products. Ultimately, the proposed PCF sensor enhances food safety monitoring and paves the way for advanced applications in the food industry, ensuring consumers receive high-quality products.

Object detection, as a key technology in computer vision, has been widely applied across various fields. However, traditional algorithms often need help with poor generalisation and low accuracy, limiting their performance in complex scenarios. With the advent of deep learning, neural networks leveraging large datasets have demonstrated remarkable improvements in generalisation and accuracy, significantly outperforming traditional methods. This study focuses on improving the YOLOv8 algorithm to address detection challenges in complex environments. The enhanced YOLOv8 model incorporates tailored modifications to its network structure, improving its feature extraction capabilities and detection efficiency. A custom vehicle dataset featuring diverse and challenging backgrounds was pre-processed and utilised for training, resulting in a robust vehicle detection model. The experimental results show that the improved YOLOv8 algorithm achieved a recall from 0.469 to 0.479 and [email protected] from 0.520 to 0.533, demonstrating significant performance gains. PyQt5-based graphical user interface was developed, providing a user-friendly platform for real-time detection and analysis. The interface allows users to input images or videos, view detection results, and adjust parameters dynamically, offering both functionality and convenience. This combination of algorithmic enhancement and intuitive interface design establishes a strong foundation for real-world applications and further advancements in multi-target detection and tracking.

Multiaccess edge computing (MEC) is a dynamic approach for addressing the capacity and ultra-latency demands caused by the pervasive growth of real-time applications in next-generation (xG) wireless communication networks. Powerful computational resource-enriched virtual machines (VMs) are used in MEC to provide outstanding solutions. However, a major challenge with using VMs in xG networks is the high overhead caused by the excessive energy demands of VMs. To address this challenge, containers, which are generally more energy-efficient and less computationally demanding, are being advocated. This paper proposes a containerised edge computing model for power optimisation in 6G-inspired massive Internet-of-Things applications. The problem is formulated as a central processing unit energy consumption cost function based on quasi-finite system observations. To achieve practicable computational complexity, an approach that uses a search heuristic based on Lyapunov techniques is employed to obtain near-optimal solutions. Important performance metrics are successfully predicted using the online look-ahead technique. The predictive model used achieves an accuracy of 97% prediction compared to actual data. To further improve resource demand, an adaptive controller is used to schedule computational resources on a time slot basis in an adaptive manner while continuing to receive workload levels to plan future resource provisioning. The proposed technique is shown to perform better compared to a competitive baseline algorithm.