Multi-objective deep reinforcement learning for computation offloading in UAV-assisted multi-access edge computing

Information Sciences, Год журнала: 2023, Номер 642, С. 119154 - 119154

Опубликована: Май 18, 2023

Язык: Английский

---

Energy Consumption Modeling and Optimization of UAV-Assisted MEC Networks Using Deep Reinforcement Learning

IEEE Sensors Journal, Год журнала: 2024, Номер 24(8), С. 13629 - 13639

Опубликована: Март 7, 2024

Unmanned aerial vehicle (UAV)-assisted multiaccess edge computing (MEC) technology has garnered significant attention and been successfully implemented in specific scenarios. The optimization of the network energy consumption relevant scenarios is essential for whole system performance due to constrained capacity UAVs. However, dynamic changes MEC resources make a challenge. In this article, multi-UAV-multiuser model established assess consumption, problem multi-UAV cooperation strategies formulated based on model. Then, multiagent deep deterministic policy gradient (MADDPG) algorithm reinforcement learning (DRL) employed resolve above problem. Each UAV equivalent single agent that cooperates with other agents train actors critic evaluation networks accomplish collaborative decision-making. addition, prioritized experience replay (PER) scheme used improve convergence training process. Simulation results show impact different by comparing algorithms. findings presented article serve as valuable reference future work optimization, specifically terms efficiency.

Язык: Английский

---

An Evolutionary Algorithm for Task Clustering and Scheduling in IoT Edge Computing

Mathematics, Год журнала: 2024, Номер 12(2), С. 281 - 281

Опубликована: Янв. 15, 2024

The Internet of Things (IoT) edge is an emerging technology sensors and devices that communicate real-time data to a network. IoT computing was introduced handle the latency concerns related cloud management, as are processed closer their point origin. Clustering scheduling tasks on considered challenging problem due diverse nature task resource characteristics. Metaheuristics optimization methods widely used in clustering scheduling. This paper new mechanism using differential evolution computing. proposed aims optimize find optimal execution times for submitted tasks. based degree similarity mechanisms use evolutionary distribute system across suitable resources. process categorizes with similar requirements then maps them appropriate To evaluate scheduling, this study conducted several simulation experiments against two established mechanisms: Firefly Algorithm (FA) Particle Swarm Optimization (PSO). configuration carefully created mimic real-world settings ensure mechanism’s applicability results’ relevance. In heavyweight workload scenario, DE started time 916.61 milliseconds, compared FA’s 1092 milliseconds PSO’s 1026.09 milliseconds. By 50th iteration, had reduced its significantly around 821.27 whereas FA PSO showed lesser improvements, at approximately 1053.06 stabilizing 956.12 results revealed outperforms regarding efficiency stability, reducing having minimal variation iterations.

Язык: Английский

---

Task offloading using GPU-based particle swarm optimization for high-performance vehicular edge computing

Journal of King Saud University - Computer and Information Sciences, Год журнала: 2022, Номер 34(10), С. 10356 - 10364

Опубликована: Ноя. 1, 2022

With mobile edge computing, vehicles can obtain nearby network resources and computability meet the growing demand for vehicular service at large scales. However, as a result of vehicle mobility offloading an extensive number tasks, congestion in wireless networks insufficient computing servers make it difficult to maintain good quality users. Moreover, access point selection is not often considered factor task execution latency. In this paper, we propose smart metaheuristic optimization model address problem low due movements limited coverage. Then, proposed used characterize overall latency by considering resource utilization, workload movement characteristics. There are two advantages our framework. First, design offers adaptive strategy automatically providing preallocation decisions with respect server state. Second, approach benefits from recent advances graphics processing unit (GPU) architectures. fact, PSO on GPU shifts process promising area terms time precision. Extensive experimental results presented demonstrate effectiveness

Язык: Английский

---

Multi-UAV computing enabling efficient clustering-based IoT for energy reduction and data transmission

Journal of Intelligent & Fuzzy Systems, Год журнала: 2023, Номер 45(1), С. 1717 - 1730

Опубликована: Май 12, 2023

Internet of Things (IoT) technologies increasingly integrate unmanned aerial vehicles (UAVs). IoT devices that are becoming more networked produce massive data. The process and memory this enormous volume data at local nodes, particularly when utilizing artificial intelligence (AI) algorithms to collect utilize useful information, have been declared vital issues. In paper, we introduce UAV computing solve greater energy consumption, delay difficulties using task offload clustered approaches, make cloud operations accessible devices. First, present a clustering technique group for transmission. After that, apply the Q-learning approach accomplish offloading allocate difficult tasks UAVs not yet fully loaded. sensor readings from CHs then collected path planning. Furthermore, We use convolutional neural network (CNN) achieve route terms coverage ratio, efficiency, motion, number packets, effectiveness current study is finally compared with existing techniques UAVs. results showed suggested strategy outperformed approaches in packets. Additionally, proposed consumed less due CNN-based planning dynamic positioning, which reduced transmits power. Overall, concluded effective improving energy-efficient responsive transmission crises.

Язык: Английский

---

Review on Meta-heuristic Algorithm-Based Priority-Aware Computation Offloading in Edge Computing System

Journal of The Institution of Engineers (India) Series B, Год журнала: 2025, Номер unknown

Опубликована: Янв. 31, 2025

Язык: Английский

---

Efficient Task Offloading and Resource Allocation in MEC

Journal of Cases on Information Technology, Год журнала: 2025, Номер 27(1), С. 1 - 22

Опубликована: Март 22, 2025

This paper proposes a novel optimization method for task offloading in Multi-Access Edge Computing (MEC) environments. The combines Ant Colony Optimization (ACO) and Genetic Algorithms (GA) to minimize total execution latency. ACO explores the solution space potential optimal solutions, while GA refines these solutions through evolutionary processes. Simulation experiments validate effectiveness of this approach, showing significant reductions overall latency compared conventional single-algorithm methods. also discusses key factors influencing strategies, providing practical insights real-world deployments. proposed hybrid ACO-GA strategy offers high-efficiency adaptable allocation problem MEC, enhancing system's performance quality.

Язык: Английский

---

Two-phase optimization modelling with swarm computation and biomimetic intelligence learning for neural network training

Computers & Industrial Engineering, Год журнала: 2025, Номер unknown, С. 111058 - 111058

Опубликована: Март 1, 2025

Язык: Английский

---

Multi-Joint Symmetric Optimization Approach for Unmanned Aerial Vehicle Assisted Edge Computing Resources in Internet of Things-Based Smart Cities

Symmetry, Год журнала: 2025, Номер 17(4), С. 574 - 574

Опубликована: Апрель 10, 2025

Smart cities are equipped with a vast number of IoT devices, which help to collect and analyze data improve the quality life for urban people by offering sustainable connected environment. However, rapid growth systems has issues related Quality Service (QoS) allocation limited resources in IoT-based smart cities. The cloud system also faces higher consumption energy extended latency. This research presents an effort overcome these challenges introducing opposition-based learning incorporated into Golden Jackal Optimization (OL-GJO) assign distributed edge capabilities diminish delay In context cities, three-layered architecture is developed, comprising system, Unmanned Aerial Vehicle (UAV)-assisted layer, layer. Moreover, controller positioned at UAV helps determine tasks. proposed approach, based on learning, put forth offer effective computing delay-sensitive multi-joint symmetric optimization uses OL-GJO, where confirms search process employed, improving task scheduling UAV-assisted computing. experimental findings exhibit that OL-GJO performs manner while offloading resources. For 200 tasks, experienced 2.95 ms, whereas Multi Particle Swarm (M-PSO) auction-based approach experience delays 7.19 ms 3.78 respectively.

Язык: Английский

---

Delay-sensitive task offloading and efficient resource allocation in intelligent edge–cloud environments: A discretized differential evolution-based approach

Applied Soft Computing, Год журнала: 2024, Номер 159, С. 111637 - 111637

Опубликована: Апрель 24, 2024

Язык: Английский

---