Scholarlink

<p>Traditional cloud computing architectures have become inadequate for supporting complex applications that demand stringent cost-efficiency and enhanced security requirements. To address these challenges, multi-access edge computing (MEC) extends cloud capabilities to the network edge, enabling localized computation and reducing latency. However, the real-time changes in task characteristics and fluctuating network conditions pose significant challenges for task offloading. To overcome these challenges, this paper proposes a low-cost MEC dynamic task offloading optimization model with privacy protection. The model adopts a three-layer cloud–edge–terminal architecture, aiming to maximize the weighted sum of privacy entropy and task processing cost by dynamically adjusting offloading strategies to optimize overall utility. The problem is formulated as a Markov Decision Process (MDP) and solved using a Hybrid Exploration Dynamic Offloading Algorithm (HEDOA), which enhances Q-learning with simulated annealing to balance exploration and exploitation more effectively. Simulation results show that the proposed offloading model significantly reduces costs and enhances privacy protection. Compared to existing algorithms, the HEDOA algorithm exhibits superior convergence and solution speed, improving the target utility by at least 10.6%.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Dynamic task offloading in MEC with joint cost reduction and privacy protection

  • 期刊论文

  • 发表日期：

    2026年3月30日

查看全文
  • Yanping Chen,
  • Yiqiang Wang,
  • Xiaomin Jin,
  • Zhongmin Wang,
  • Chen Lu

摘要

Traditional cloud computing architectures have become inadequate for supporting complex applications that demand stringent cost-efficiency and enhanced security requirements. To address these challenges, multi-access edge computing (MEC) extends cloud capabilities to the network edge, enabling localized computation and reducing latency. However, the real-time changes in task characteristics and fluctuating network conditions pose significant challenges for task offloading. To overcome these challenges, this paper proposes a low-cost MEC dynamic task offloading optimization model with privacy protection. The model adopts a three-layer cloud–edge–terminal architecture, aiming to maximize the weighted sum of privacy entropy and task processing cost by dynamically adjusting offloading strategies to optimize overall utility. The problem is formulated as a Markov Decision Process (MDP) and solved using a Hybrid Exploration Dynamic Offloading Algorithm (HEDOA), which enhances Q-learning with simulated annealing to balance exploration and exploitation more effectively. Simulation results show that the proposed offloading model significantly reduces costs and enhances privacy protection. Compared to existing algorithms, the HEDOA algorithm exhibits superior convergence and solution speed, improving the target utility by at least 10.6%.