Scholarlink

<p>One of the challenges in the 5G-enabled industrial Internet of Things (IIoT) systems is efficient task offloading. As this problem is a type of NP-hard problems, it can be solved with optimization algorithms. As well, in this optimization problem, it must be considered that low-latency execution, device-side energy consumption, and balanced use of heterogeneous Mobile Edge Computing (MEC) servers simultaneously. In this study, the IIoT task offloading problem is solved by the scalarized multi-criteria fitness function. This formulation distinguished the local and MEC servers. As well, it uses a Shannon-inspired uplink model to evaluate device-side computation and communication energy and measures MEC load imbalance through capacity-aware workload rather than task counts. One of the well-known metaheuristic algorithms that are able to solve optimization problems is sand cat swarm optimization (SCSO) problems due to their characteristic of solving continuous problems. In this study the continuous SCSO transfers into a discrete decision process through a sigmoid transfer function and integration of chaotic maps. This method updates the search agent’s position to improve exploration and exploitation besides avoiding trapping a local optimum. The proposed framework is evaluated over 30 independent runs under three benchmark scenarios and compared with PSO, GWO, SCSO, d-SCSO. The results show that d-chSCSO-WA achieves the best observed mean fitness in Cases 1 and 2, improving over the strongest non-proposed comparator by 21.8% and 14.1%, respectively. In the largest Case 3, the Bernoulli variant ranks second, remaining within 5.2% of AGA-SAPSO-TO while outperforming PSO, MSSA-TO, and SCSO by 13.9%, 27.9%, and 45.3%, respectively. The obtained results analysis further confirm that chaotic control and workload-aware modification significantly improve fitness, stability, and load distribution under nominal and stressed IIoT operating conditions.</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

Enhancing IIoT Task Offloading Using Workload-Aware Discrete Chaotic Sand Cat Swarm Optimization

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月30日

查看全文
  • Amir Seyyedabbasi,
  • Ali Asghar Pour Haji Kazem,
  • Neda Dadashkhani,
  • Ahmet Gurhanli,
  • Jawad Rasheed

摘要

One of the challenges in the 5G-enabled industrial Internet of Things (IIoT) systems is efficient task offloading. As this problem is a type of NP-hard problems, it can be solved with optimization algorithms. As well, in this optimization problem, it must be considered that low-latency execution, device-side energy consumption, and balanced use of heterogeneous Mobile Edge Computing (MEC) servers simultaneously. In this study, the IIoT task offloading problem is solved by the scalarized multi-criteria fitness function. This formulation distinguished the local and MEC servers. As well, it uses a Shannon-inspired uplink model to evaluate device-side computation and communication energy and measures MEC load imbalance through capacity-aware workload rather than task counts. One of the well-known metaheuristic algorithms that are able to solve optimization problems is sand cat swarm optimization (SCSO) problems due to their characteristic of solving continuous problems. In this study the continuous SCSO transfers into a discrete decision process through a sigmoid transfer function and integration of chaotic maps. This method updates the search agent’s position to improve exploration and exploitation besides avoiding trapping a local optimum. The proposed framework is evaluated over 30 independent runs under three benchmark scenarios and compared with PSO, GWO, SCSO, d-SCSO. The results show that d-chSCSO-WA achieves the best observed mean fitness in Cases 1 and 2, improving over the strongest non-proposed comparator by 21.8% and 14.1%, respectively. In the largest Case 3, the Bernoulli variant ranks second, remaining within 5.2% of AGA-SAPSO-TO while outperforming PSO, MSSA-TO, and SCSO by 13.9%, 27.9%, and 45.3%, respectively. The obtained results analysis further confirm that chaotic control and workload-aware modification significantly improve fitness, stability, and load distribution under nominal and stressed IIoT operating conditions.