Search

The deployment of increasingly diverse services on edge devices is becoming increasingly prevalent. Efficiently deploying functionally heterogeneous services to resource heterogeneous edge nodes while achieving superior user experience is a challenge that every edge system must address. In this paper, we propose a container-node graph (CN-Graph)-based container optimal matching deployment algorithm, edge Kuhn-Munkres algorithm (EKM) based on container-node graph, designed for heterogeneous environment to optimize system performance. Initially, containers are categorized by functional labels, followed by construction of a CN-Graph model based on the relationship between containers and nodes. Finally, the container deployment problem is transformed into a weighted bipartite graph optimal matching problem. In comparison with the mainstream container deployment algorithms, Swarm, Kubernetes, and the recently emerged ECSched-dp algorithm, the EKM algorithm demonstrates the ability to effectively enhance the average runtime performance of containers to 3.74 times, 4.10 times, and 2.39 times, respectively. [ABSTRACT FROM AUTHOR]

The rapid development of Internet of Vehicles (IoV) technology has led to a sharp increase in vehicle data. Traditional cloud computing is no longer sufficient to meet the high bandwidth and low latency requirements of IoV tasks. Ensuring the service quality of applications on in-vehicle devices has become challenging. Edge computing technology moves computing tasks from the cloud to edge servers with sufficient computing resources, effectively reducing network congestion and data propagation latency. The integration of edge computing and IoV technology is an effective approach to realizing intelligent applications in IoV.This paper investigates the deployment of vehicle edge servers in cloud-edge collaborative environment. Taking into consideration the vehicular mobility and the computational demands of IoV applications, the vehicular edge server deployment within the cloud-edge collaborative framework is formulated as a multi-objective optimization problem. This problem aims to achieve two primary objectives: minimizing service access latency and balancing server workload. To address this problem, a model is established for optimizing the deployment of vehicle edge servers and a deployment approach named VSPR is proposed. This method integrates hierarchical clustering and reinforcement learning techniques to effectively achieve the desired multi-objective optimization. Experiments are conducted using a real datasets from Shanghai Telecom to comprehensively evaluate the performance of workload balance and service access latency of vehicle edge servers under different deploy methods. Experimental results demonstrate that VSPR achieves an optimized balance between low latency and workload balancing while ensuring service quality, and outperforms SRL, CQP, K-means and Random algorithm by 4.76%, 44.59%, 40.78% and 69.33%, respectively. [ABSTRACT FROM AUTHOR]

Given the problems with a centralized cloud and the emergence of ultra-low latency applications, and the needs of the Internet of Things (IoT), it has been found that novel methods are needed to support centralized cloud technology. Mobile edge computing is one of the solutions to mitigate these challenges. In this paper, we study task caching at Device to Device (D2D)-assisted network edge. In the proposed scheme, we predict the possibility of re-requesting tasks in the future using convolutional neural networks (CNN). Based on this predicted possibility, the number of last requests, and the number of cached versions of this task type in the neighbors, in addition to the characteristics of a task itself, including the required cache volume and processing resources, we rank the tasks using the proposed Multi-Criteria Task Ranking using Predicted requests (MCTRP) scheme and select the best replacement option in the cache of each Mobile User Equipment (MUE). The proposed scheme has proved to be of considerable benefit in terms of reducing delay and energy consumption and improving the utility of MUEs. [ABSTRACT FROM AUTHOR]

The rapid integration of cloud computing and edge computing has brought the cloud-edge environment into the spotlight in information technology. Within this context, the selection of high-quality and reliable services is crucial to meet the needs of users. However, ensuring the reliability of service information is a challenge due to its vulnerability to tampering. This research paper proposes a method for service selection in the cloud-edge environment based on blockchain smart contracts. By leveraging blockchain technology, this method achieves decentralized and trustworthy service selection. Through smart contracts, user interactions are securely recorded, significantly reducing the risk of information tampering and enhancing information reliability. Additionally, the Arithmetic Optimization Algorithm is improved for service selection on the blockchain by introducing mutation and crossover operations. Experimental results demonstrate that this method effectively prevents tampering with service information and improves the utility value of selected services compared to traditional methods and metaheuristic algorithms mentioned. [ABSTRACT FROM AUTHOR]

Federated learning (FL) is a decentralized approach to training machine learning model. In the traditional architecture, the training requires getting the whole data what causes a threat to the privacy of the sensitive data. FL was proposed to overcome the cited limits. The principal of FL revolves around training machine learning models locally on individual devices instead of gathering all the data in a central server, and only the updated models are shared and aggregated. Concerning e-learning, it is about using electronic/digital technology to deliver educational content in order to facilitate the learning. It becomes popular with the advancement of the internet and digital devices mainly after the COVID-19. In this work, we propose an e-learning recommendation system based on FL architecture where we can propose suitable courses to the learner. Because of the important number of connected learners looking for online courses, the FL encounters a problem: bottleneck communication. This situation can cause the increase of the computational load, the longer time of the aggregation, the saturation of the resources, etc. As solution, we propose using the edge computing potentials so that the aggregation will be performed first in the edge layer then in the central server, reducing hence, the need for continuous data transmission to the server and enabling a faster inference while keeping the security and privacy of the data. The experiments carried out prove the effectiveness of our approach in solving the problem addressed in this work. [ABSTRACT FROM AUTHOR]

Summary: In sedimentology, the integration of intelligent engineering decision‐making with edge computing environments aims to furnish engineers and decision‐makers with precise, real‐time insights into sediment‐related issues. This approach markedly reduces data transfer time and response latency by harnessing the computational power of edge computing, thereby bolstering the decision‐making process. Concurrently, the establishment of a sediment knowledge graph serves as a pivotal conduit for disseminating sediment‐related knowledge in the realm of intelligent engineering decision‐making. Moreover, it facilitates a comprehensive exploration of the intricate evolutionary and transformative processes inherent in sediment materials. By unveiling the evolutionary trajectory of life on Earth, the sediment knowledge graph catalyzes a deeper understanding of our planet's history and dynamics. Relationship extraction, as a key step in knowledge graph construction, implements automatic extraction and establishment of associations between entities from a large amount of sedimentary literature data. However, sedimentological literature presents multi‐source heterogeneous features, which leads to a weak representation of hidden relationships, thus decreasing the accuracy of relationship extraction. In this article, we propose an attribute‐extended relation extraction approach (AERE), which is specifically designed for sedimentary relation extraction scenarios. First, context statements containing sediment entities are obtained from the literature. Then, a cohesive hierarchical clustering algorithm is used to extend the relationship attributes between sediments. Finally, mine the relationships between entities based on AERE. The experimental results show that the proposed model can effectively extract the hidden relations and exhibits strong robustness in dealing with redundant noise before and after sentences, which in turn improves the completeness of the relations between deposits. After the relationship extraction, a proprietary sediment knowledge graph is constructed with the extracted triads. [ABSTRACT FROM AUTHOR]

In the rapidly advancing information era, various human behaviors are being precisely recorded in the form of data, including identity information, criminal records, and communication data. Law enforcement agencies can effectively maintain social security and precisely combat criminal activities by analyzing the aforementioned data. In comparison to traditional data analysis methods, deep learning models, relying on the robust computational power in cloud centers, exhibit higher accuracy in extracting data features and inferring data. However, within the architecture of cloud centers, the transmission of data from end devices introduces significant latency, hindering real-time inference of data. Furthermore, low-latency edge computing architectures face limitations in direct deployment due to relatively weak computing and storage capacities of nodes. To address these challenges, a lightweight distributed knowledge graph completion architecture is proposed. Firstly, we introduce a lightweight distributed knowledge graph completion architecture that utilizes knowledge graph embedding for data analysis. Subsequently, to filter out substandard data, a personnel data quality assessment method named PDQA is proposed. Lastly, we present a model pruning algorithm that significantly reduces the model size while maximizing performance, enabling lightweight deployment. In experiments, we compare the effects of 11 advanced models on completing the knowledge graph of public security personnel information. The results indicate that the RotatE model outperforms other models significantly in knowledge graph completion, with the pruned model size reduced by 70%, and hits@10 reaching 86.97%. [ABSTRACT FROM AUTHOR]

Energy consumption is a vital issue when optimum usage and carbon footprint are all considered in today's Internet of Things (IoT) environments. Considering edge computing, that becomes too critical in terms of wireless devices with limited battery power. Especially in healthcare applications, the defined IoHT approach requires sustainability while future massive solutions may result negative outputs in terms of carbon footprint. So, optimum energy consumption seems positive in terms of multiple ways. In the literature, one trendy method is using clustering for lowering the energy consumption within the Internet of Healthcare Things (IoHT) environment on edge computing. In this study, optimization of energy consumption in IoHT was done via improved Genetic Electro-Search Optimization (GESO) algorithm. According to the obtained findings in the performed applications, GESO was effective enough in finding optimum conditions of energy consumption for an active IoHT setup. [ABSTRACT FROM AUTHOR]

Edge computing is a new paradigm for processing data at the edge of networks. There are a variety of edge computing scenarios, depending on the situation. In this paper, we investigate an architecture with heterogenous devices for intelligence aquaculture. The system will collect water sensor data and run real-time video-based fish detection with a Deep Learning algorithm and Deepstream. Each system was monitored to ensure all the architecture design was running properly. Kubernetes Lightweight Kubernetes (K3s) was used to manage all applications deployed in Docker containers. In addition, to synchronizing the heterogenous devices and visualizing the node's resource system, the Rancher Kubernetes Engine is used to coordinate its resources. The container-based architecture with embedded GPU can work properly for fish detection and the POD schedule scheme that we propose show an improvement in container AI performance and increased around 20%. The architecture of this system can be referenced as a model for edge computing ecosystems. [ABSTRACT FROM AUTHOR]

Misalignment, imbalance, induced vibrations, and noise in rotating machines must be identified early on using condition monitoring and signal processing techniques. If it is not detected early, the machine's reliability will suffer, potentially resulting in a catastrophic failure of the machine components. In this study, a web application for real-time fault detection is designed and built using a novel approach of edge computing and IoT. Vibration signature analysis are used to determine the severity of faults in machine rotating components and to provide an early warning even when the maintenance crew is located in a remote location. The vibration spectrum analysis results are successfully obtained and verified using the vibration-metre tool VibXpert-II. The purpose of this research is to improve real-time condition monitoring of rotating systems like bearings and induction motors and make it available on an online platform for predictive maintenance. Vibration signatures provide more accurate information regarding the type and location of rotor faults than current signatures. [ABSTRACT FROM AUTHOR]

Thermographic inspection is particularly effective in identifying thermal bridges because it visualizes temperature differences on the building's surface. The focus of this work is on energy-efficient computing for deep learning-based thermal bridge (anomaly) detection models. In this study, we concentrate on object detection-based models such as Mask R-CNN_FPN_50, Swin-T Transformer, and FSAF. We do benchmark tests on TBRR dataset with varying input sizes. To overcome the energy-efficient design, we apply optimizations such as compression, latency reduction, and pruning to these models. After our proposed improvements, the inference of the anomaly detection model, Mask R-CNN_FPN_50 with compression technique, is approximately 7.5% faster than the original. Also, more acceleration is observed in all models with increasing input size. Another criterion we focus on is total energy gain for optimized models. Swin-T transformer has the most inference energy gains for all input sizes (≈ 27 J for 3000 x 4000 and ≈ 14 J for 2400 x 3400). In conclusion, our study presents an optimization of size, weight, and power for vision-based anomaly detection for buildings. [ABSTRACT FROM AUTHOR]

The recent widespread of AI-powered real-time applications necessitates the use of edge computing for inference task offloading. Power constrained edge devices are required to balance between processing inference tasks locally or offload to edge servers. This decision is determined according to the time constraint demanded by the real-time nature of applications, and the energy constraint dictated by the device's power budget. This problem is further exacerbated in the case of systems leveraging multiple local inference models varying in size and accuracy. In this work, we tackle the problem of assigning inference models to inference tasks either using local inference models or by offloading to edge servers under time and energy constraints while maximizing the overall accuracy of the system. This problem is shown to be strongly NP-hard and therefore, we propose a hybrid genetic algorithm (HGSTO) to solve this problem. We leverage the speed of simulated annealing (SA) with the accuracy of genetic algorithms (GA) to develop a hybrid, fast and accurate algorithm compared with classic GA, SA and Particle Swarm Optimization (PSO). Experiment results show that HGSTO achieved on-par or higher accuracy than GA while resulting in significantly lower scheduling times compared to other schemes. [ABSTRACT FROM AUTHOR]

Optimizing computational offloading in Mobile Edge Computing (MEC) environments presents a multifaceted challenge requiring innovative solutions. Soft computing, recognized for its ability to manage uncertainty and complexity, emerges as a promising approach for addressing the dynamic multi-objective evaluation inherent in computational offloading scenarios. This paper conducts a comprehensive review and analysis of soft computing approaches for Dynamic Multi-Objective Evaluation of Computational Offloading (DMOECO), aiming to identify trends, analyze existing literature, and offer insights for future research directions. Employing a systematic literature review (SLR) methodology, we meticulously scrutinize 50 research articles and scholarly publications spanning from 2016 to November 2023. Our review synthesizes advancements in soft computing techniques, including fuzzy logic, neural networks, evolutionary algorithms, and probabilistic reasoning, as applied to computational offloading optimization within MEC environments. Within this comprehensive review, existing approaches are categorized and analyzed into distinct research lines based on methodologies, objectives, evaluation metrics, and application domains. The evolution of soft computing-based DMOECO strategies is emphasized, showcasing their effectiveness in dynamically balancing various computational objectives, including energy consumption, latency, throughput, user experience, and other pertinent factors in computational offloading scenarios. Key challenges, including scalability issues, lack of real-world deployment validation, and the need for standardized evaluation benchmarks, are identified. Insights and recommendations are provided to enhance computational offloading optimization. Furthermore, collaborative efforts between academia and industry are advocated to bridge the theoretical developments with practical implementations. This study pioneers the use of SLR methodology, offering valuable perspectives on soft computing in DMOECO and synthesizing state-of-the-art approaches. It serves as a crucial resource for researchers, practitioners, and stakeholders in the MEC domain, illuminating trends and fostering continued innovation in computational offloading strategies. [ABSTRACT FROM AUTHOR]

The increased data transmission and number of devices involved in communications among distributed systems make it challenging yet significantly necessary to have an efficient and reliable networking middleware. In robotics and autonomous systems, the wide application of ROS 2 brings the possibility of utilizing various networking middlewares together with DDS in ROS 2 for better communication among edge devices or between edge devices and the cloud. However, there is a lack of comprehensive communication performance comparison of integrating these networking middlewares with ROS 2. In this study, we provide a quantitative analysis for the communication performance of utilized networking middlewares including MQTT and Zenoh alongside DDS in ROS 2 among a multiple host system. For a complete and reliable comparison, we calculate the latency and throughput of these middlewares by sending distinct amounts and types of data through different network setups including Ethernet, Wi-Fi, and 4G. To further extend the evaluation to real-world application scenarios, we assess the drift error (the position changes) over time caused by these networking middlewares with the robot moving in an identical square-shaped path. Our results show that CycloneDDS performs better under Ethernet while Zenoh performs better under Wi-Fi and 4G. In the actual robot test, the robot moving trajectory drift error over time (96 s) via Zenoh is the smallest. It is worth noting we have a discussion of the CPU utilization of these networking middlewares and the perfosrmance impact caused by enabling the security feature in ROS 2 at the end of the paper. [ABSTRACT FROM AUTHOR]

Effective traffic management is critical in the modern age of smart cities to guarantee seamless transit, lessen traffic, and protect the environment. Therefore, to enhance road traffic monitoring, this research presents a novel Quantum Optimized LSTM (QO-LSTM) framework that leverages Quantum Machine Learning techniques with Long Short-Term Memory in an edge cloud environment. The quantum circuit is used to improve predictions via quantum-enhanced optimization, while the LSTM network is utilized to extract temporal relationships from traffic data. The use of this hybrid approach in edge cloud infrastructure offers low latency and great scalability, making it ideal for real-time applications in smart city contexts. The QO-LSTM model's performance was assessed using several measures, yielding results such as a 99.32% Coefficient of Determination (R2), a 1.96% Root Mean Squared Error (RMSE), and a 0.97% Mean Absolute Error (MAE) which are far better when compared with other models like GRU, LSTM and SAE. Additionally, the model showed great prediction accuracy and reliability with an Explained Variance Score (EVS) of 99.33% and a Mean Absolute Percentage Error (MAPE) of 1.07%. Traffic peaks were also identified followed by the peak durations to gain an understanding of traffic congestion patterns. Moreover, by integrating these innovations, the findings reveal that the model considerably improves the accuracy and responsiveness of traffic predictions, allowing for more effective traffic management approaches and real-time decision-making. [ABSTRACT FROM AUTHOR]

With the rapid advancement in technology, numerous advanced vehicular applications have emerged that generate large volumes of data that need to be processed on the fly. The vehicles' computing resources are limited and constrained in processing the huge amount of data generated by these applications. Cloud data centers, which are large and capable of processing the generated data, tend to be far away from the vehicles. The long distance between the cloud and the vehicles results in large transmission delays, making the cloud less suitable for executing such data. To address the long-standing issue of huge transmission delays in the cloud, edge computing, which deploys computing servers at the edge of the network, was introduced. The edge computing network shortens the communication distance between the vehicles and the processing resources and also provides more powerful computation compared to the vehicles' computing resources. The advantages offered by the vehicular edge network can only be fully realized with robust and efficient resource allocation. Poor allocation of these resources can lead to a worse situation than the cloud. In this paper, a hybrid Marine Predatory and Particle Swarm Optimization Algorithm (MPA–PSO) is proposed for optimal resource allocation. The MPA–PSO algorithm takes advantage of the effectiveness and reliability of the global and local search abilities of the Particle Swarm Optimization Algorithm (PSO) to improve the suboptimal global search ability of the MPA. This enhances the other steps in the MPA to ensure an optimal solution. The proposed MPA–PSO algorithm was implemented using MATLAB alongside the conventional PSO and MPA, and the proposed MPA–PSO recorded a significant improvement over the PSO and MPA. [ABSTRACT FROM AUTHOR]

This paper focuses on the smart security aspect of smart city initiatives, and specifically on road traffic monitoring. We describe the design and deployment of smart traffic monitoring in a pilot project located in the Selangor Cyber Valley, Selangor, Malaysia. Live roadside closed-circuit television footages are streamed to a remote network video recorder (NVR) and video management system. Videos from the NVR are also transmitted to a workstation at the Centre of Excellence for Advanced Sensor Technology for training, analysis, and inference with an artificial intelligence model. The proposed architecture allows for model updates without disruption to the running of the system. YOLOv7 is used for the object detection task of vehicle type identification, with enhancements for vehicle counting and brand/manufacturer classification. Identifying vehicles is crucial in regard to managing traffic flow, scheduling the movements of heavy vehicles, and identifying infrastructure improvements to alleviate congestion. The proposed model achieves precision, recall, and mean average precision (mAP) values of 80.6%, 88.3%, and 87.8%, respectively, surpassing the YOLOv7 model with h5 configuration by 3.6%, 9.3%, and 6.8%. The proposed model accurately detects 11 classes with mAP values of above 80%, except for buses, where the mAP is 70.6%. The similar characteristics of buses and vans (which are small objects, as they are captured from long distances) contribute to this lower accuracy, suggesting the need for more images and augmentation techniques to improve detection. The challenges encountered during and after deployment are addressed, and insights and recommendations are presented for future implementations. [ABSTRACT FROM AUTHOR]

Federated learning (FL) is a novel approach to privacy‐preserving machine learning, enabling remote devices to collaborate on model training without exchanging data among clients. However, it faces several challenges, including limited client‐side processing capabilities and non‐IID data distributions. To address these challenges, we propose a partitioned FL architecture that a large CNN is divided into smaller networks, which train concurrently with other clients. Within a cluster, multiple clients concurrently train the ensemble model. The Jensen–Shannon divergence quantifies the similarity of predictions across submodels. To address discrepancies in model parameters between local and global models caused by data distribution, we propose an ensemble learning method that integrates a penalty term into the local model's loss calculation, thereby ensuring synchronization. This method amalgamates predictions and losses across multiple submodels, effectively mitigating accuracy loss during the integration process. Extensive experiments with various Dirichlet parameters demonstrate that our system achieves accelerated convergence and enhanced performance on the CIFAR‐10 and CIFAR‐100 image classification tasks while remaining robust to partial participation, diverse datasets, and numerous clients. On the CIFAR‐10 dataset, our method outperforms FedAvg, FedProx, and SplitFed by 6%–8%; in contrast, it outperforms them by 12%–18% on CIFAR‐100. [ABSTRACT FROM AUTHOR]

In the past few years, with the increase in population and health concerns, there has been a need for efficient health monitoring solutions that can help patients monitor their health consistently to be aware of any health risks at the initial stage. The advancement in sensing and smart technologies helps monitor human behaviors to predict health risks. In this work, a dynamic decision-based activity prediction system is proposed using Random Forest, SVM, Decision Trees, Long Short-Term Memory (LSTM), and Gated Recurrent Unit (GRU) on an edge device. We train the models using features from the MHealth dataset, such as acceleration, rate of turn, and magnetic field, to predict activities such as standing, climbing, running, and jogging, collected from various sensors. Our framework dynamically selects between machine learning (ML) and deep learning (DL) algorithms based on real-time data size and edge device capabilities, ensuring optimal performance and resource utilization. The results for the proposed models are compared and analyzed. The experimental results indicate that among all machine learning methods, Random Forest achieves the highest overall accuracy at 98%, while in deep learning algorithms, both LSTM and GRU reach a maximum accuracy of 98%. [ABSTRACT FROM AUTHOR]

ABSTRACT Digital textbooks (DTs) have evolved from DT 1.0, which simply converted paper textbooks to PDF format, to DT 2.0, which provides various multimedia content, for example, video and audio content. DTs have now advanced to DT 3.0, which enhances learner engagement through gamification and simulations. Recently, with the advancement of cloud computing technology and digital devices, for example, tablets, DT 4.0, which supports personalised learning through artificial intelligence (AI) tutors and chatbots, has been realised. South Korea is actively implementing a policy to distribute artificial intelligence–based DTs, equivalent to DT 4.0, to all schools under national leadership. For artificial intelligence–based DTs (AIDTs) in South Korea to develop into a sustainable education system, reliance on cloud computing alone is insufficient. It is also necessary to build layers of fog computing and edge computing from the initial stage. There are concerns that AIDTs may exacerbate the learning gap because they are more likely to be utilised actively by high‐performing students with established self‐directed learning habits rather than struggling students. Thus, it is essential to enhance usage monitoring and explore strategies that provide educational scaffolding to prevent differences in the level of AIDT utilisation from leading to a widening learning gap. [ABSTRACT FROM AUTHOR]

In the rapidly growing Internet of Things (IoT) landscape, federated learning (FL) plays a crucial role in enhancing the performance of heterogeneous edge computing environments due to its scalability, robustness, and low energy consumption. However, one of the major challenges in such environments is the efficient selection of edge nodes and the optimization of resource allocation, especially in dynamic and resource-constrained settings. To address this, we propose a novel architecture called Multi-Edge Clustered and Edge AI Heterogeneous Federated Learning (MEC-AI HetFL), which leverages multi-edge clustering and AI-driven node communication. This architecture enables edge AI nodes to collaborate, dynamically selecting significant nodes and optimizing global learning tasks with low complexity. Compared to existing solutions like EdgeFed, FedSA, FedMP, and H-DDPG, MEC-AI HetFL improves resource allocation, quality score, and learning accuracy, offering up to 5 times better performance in heterogeneous and distributed environments. The solution is validated through simulations and network traffic tests, demonstrating its ability to address the key challenges in IoT edge computing deployments. [ABSTRACT FROM AUTHOR]

ABSTRACT Mobile edge computing (MEC) facilitates storage, cloud computing, and analysis capabilities near to the users in 5G communication systems. MEC and deep learning (DL) are combined in 5G networks to enable automated network management that provides resource allocation (RA), energy efficiency (EE), and adaptive security, thereby reducing computational costs and enhancing user services. A hybrid quantum‐classical convolutional neural network (HQCCNN) with simplicial attention network (SAN) is presented in the study that allocates appropriate resources for various users in the network. First, the green anaconda optimization (GAO) algorithm is used to optimize the objective function for effective RA. Consequently, the neural network receives the optimized objective functions to allocate resources. In the study, the suggested HQCCNN‐GAO model assesses the degree of need for every user and, based on those needs, allots resources to every user in the 5G network while preserving higher throughput and EE. Throughput, latency, mean square errors, processing time, bit error rates, and EE are used to measure the proposed model's efficiency. A few of the RA models that are now in use are contrasted with the outcomes of the suggested method. From the obtained outcomes, it is noticed that the suggested model provides a low latency of 0.08 s and a high throughput of 790 kbps for a range of network users. [ABSTRACT FROM AUTHOR]

The rapid expansion of the Internet of Things (IoT) necessitates advanced routing schemes capable of meeting the stringent demands for low latency and high accuracy, which are critical for applications such as autonomous vehicles and telemedicine. Traditional edge computing methods often struggle with elevated latency, rendering them unsuitable for time-sensitive applications. Additionally, many reinforcement learning (RL) algorithms require action space discretization, which can introduce biases and dimensionality challenges. This paper introduces a novel Soft Actor-Critic (SAC)-based distributed routing scheme for edge computing, specifically designed to address these limitations. By integrating RL with Maximum Entropy principles and employing a decentralized approach, the proposed model enhances network performance, reduces delays, and effectively manages multi-optimality criteria. The distributed routing scheme operates independently of a centralized controller, allowing routers to make autonomous decisions and adapt seamlessly to changes in the network. This is accomplished through a Markov Decision Process (MDP) that optimizes routing paths based on various factors, including node depth, energy consumption, and transmission probability. The methodology encompasses local training phases for individual nodes, followed by federated training to refine the model across the network. Experimental results conducted on topologies of varying scales demonstrate the model’s efficacy in achieving high accuracy and efficient convergence, particularly in dynamic IoT environments. These findings underscore the potential of the proposed SAC-based distributed routing scheme as a robust solution for enhancing routing efficiency and reliability in the evolving landscape of IoT applications.IMPACT STATEMENTThe rapid expansion of Internet of Things (IoT) applications demands advanced routing solutions to ensure low latency and high accuracy, crucial for sectors like autonomous vehicles and telemedicine. Traditional edge computing methods struggle with elevated latency, while many reinforcement learning (RL) algorithms face challenges with action space discretization, leading to biases and dimensionality issues. This study introduces a novel Soft Actor-Critic (SAC)-based distributed routing scheme to address these limitations. Integrating Maximum Entropy principles with RL enhances exploration and decision-making stability. The decentralized approach allows routers to make autonomous, real-time decisions based on local network conditions, optimizing routing paths through a Markov Decision Process (MDP). Experimental results from various simulated IoT network topologies show the model’s superior performance in reducing delays and maintaining bandwidth. This research paves the way for more reliable, low-latency IoT applications, significantly enhancing routing efficiency and network adaptability in dynamic IoT environments. [ABSTRACT FROM AUTHOR]

Attacks on IoT devices are increasing day by day. Since IoT devices nowadays have become an integral part of our daily lives, the data gathered from IoT devices benefits intruders in many ways. Financial and Healthcare institutions also allow their customers to use their services by using handheld IoT devices. Smart homes and autonomous vehicles use many IoT devices to gather data through the built-in sensors and send it to the Edge server for further processing. The computation result on the Edge server determines the decision to fulfill the user-specific needs. As these data are vital in the future cycle of execution of an intelligent algorithm of IoT device software program, hence the data are not just of temporary use, but it is transferred to a Cloud server for permanent storage. The data flows from IoT sensors to the Edge server, then from the Edge server to the Cloud server. Here the riskiest part for data to stay is on the Edge server. To counter such a security risk, we proposed and implemented the Adaptive Epsilon Greedy Reinforcement Learning (AEGRL) method which is the extension of the traditional Epsilon (ℇ) greedy reinforcement learning method. The proposed method works efficiently for both static and dynamic environments. Experimental results show that our proposed security method outperforms the recent similar security approaches in terms of scalability, robustness, and accuracy. Article Highlights: The research work proposed in this paper is to prevent malicious attacks on the IoT edge server since the edge server continuously gathers data from the surrounding IoT devices. The following are the main highlights of the paper, which bring novelty to our research work. We have considered the dynamic nature of traffic data concerning volume and the pattern of malicious data. The metadata of the malicious packets keeps changing by the smart hackers who also use intelligent tools to cross the common security barriers. Considering the dynamic environment over the Edge server we extended the traditional reinforcement learning approach to include Adaptive ability on the dynamic constraints. We experimented with both simulation-based and real-time environments. [ABSTRACT FROM AUTHOR]

Collaborative edge and cloud computing is a promising computing paradigm for reducing the task response delay and energy consumption of devices. In this paper, we aim to jointly optimize task offloading strategy, power control for devices, and resource allocation for edge servers within a collaborative device‐edge‐cloud computing system. We formulate this problem as a constrained multiobjective optimization problem and propose a joint optimization algorithm (JO‐DEC) based on a multiobjective evolutionary algorithm to solve it. To address the tight coupling of the variables and the high‐dimensional decision space, we propose a decoupling encoding strategy (DES) and a boundary point sampling strategy (BPS) to improve the performance of the algorithm. The DES is utilized to decouple the correlations among decision variables, and BPS is employed to enhance the convergence speed and population diversity of the algorithm. Simulation results demonstrate that JO‐DEC outperforms three state‐of‐the‐art algorithms in terms of convergence and diversity, enabling it to achieve a smaller task response delay and lower energy consumption. [ABSTRACT FROM AUTHOR]

IoT devices generate a massive amount of sensitive data, which is transferred tremendously to the cloud for processing and decision-making. The most significant issues that need to be solved for IoT devices are improving energy efficiency and guaranteeing data security while several devices are connected. In this paper, for edge computing, a hybrid algorithm is proposed that uses compression and encryption in the same manner to improve efficiency in terms of energy and data security in IoT devices. The authenticated encryption with associated data (AEAD) ChaCha12-Poly1305 algorithm and improved SZ 2.1 compression are used in this hybrid architecture. While sending the data to the edge, confidentiality, integrity, and authentication are maintained. Several experiments were conducted considering the driver stress detection dataset (Bernstein DJ. Lecture Notes in Computer Science including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics 4986:84–97 (2008); Lakshminarasimhan S et al. LNCS 6852(1):366–379 (2011)) and gas-sensor dataset (Ibarria L et al. Comput Graph Forum 22(3):343–348 (2003)) for monitoring the home activity to validate this approach. The performance of the proposed improved SZ 2.1 compression algorithm is compared with the five baseline algorithms including SZ 1.4, original SZ 2.1, selective compression, algorithm-based fault tolerance (ABFT), and digit rounding algorithms. The key parameters used in the experiment to measure the performance of the proposed algorithm are data reduction, compression ratio, power consumption, encryption time, total processing time, and error. Using the improved SZ 2.1 compression algorithm in conjunction with the ChaCha12-Poly1305 AEAD algorithm, the device's battery life is also enhanced by about 10% while ensuring the security of data disseminated to the Edge. The use of the proposed secure hybrid model reduces both encryption and overall processing time by 95% and 98% respectively for both datasets. [ABSTRACT FROM AUTHOR]

ABSTRACT The wireless intelligent computing paradigm has significantly provided services to various sectors in today's technology‐driven landscape. Despite its popularity, wireless intelligent computing faces challenges in addressing time‐sensitive tasks due to the physical distance between servers from users. Edge computing has been introduced for the internet of things (IoT) as an effective complement to enhance the wireless intelligent computing capacity for handling latency‐critical tasks. However, the limited resources of IoT and edge nodes can lead to suboptimal task management. In response to these challenges, we propose a lightweight approach that leverages a hybrid technique combining the whale optimization algorithm (WOA) with adaptive inertia weight and a genetic algorithm component. This method aims to enhance the efficiency of task offloading in a cloud‐edge computing environment. Experimental results demonstrate that the proposed strategy not only addresses the limitations of traditional methods but also achieves significant improvements, a 34% increase in makespan minimization, an 11% reduction in task rejection ratio, a 17% decrease in execution cost, and a 15% improvement in energy utilization compared to WOAs. The simulation results highlight the effectiveness of the proposed hybrid algorithm in enhancing quality of service (QoS) metrics for latency‐sensitive IoT applications. [ABSTRACT FROM AUTHOR]

Owing to the rapid increase in electric vehicle integration and the uncoordinated charging behaviour of electric vehicles, the overloading risk of distribution transformers has deteriorated. This impact caused by large‐scale electric vehicle integration can be effectively reduced through the orderly guidance of electric vehicle charging behaviours. Here, a dispatching strategy for charging loads is proposed to address the problems of the uncoordinated charging demand in electric vehicles and overloading risk of distribution transformers in residential areas. First, an edge‐side dynamic index of the electric vehicle hosting capacity is proposed to guide the optimal dispatching of charging loads. Subsequently, an optimal dispatching model of the charging loads is established based on edge computing. The edge‐side dispatching strategy for the charging loads is then further updated considering the participation willingness of electric vehicle users. Finally, the effectiveness of the proposed control strategy is validated using a modified residential distribution network in Tianjin. The results show that the proposed strategy can effectively decrease the overloading risk of distribution transformers while realizing the efficient operation of electric vehicles on the edge side. [ABSTRACT FROM AUTHOR]