.png){kind=icon}
.png){kind=icon}
Return
Membrane computing for IoT task offloading: An efficient multi-objective constrained optimization framework
.png){kind=icon}
PRE
AI
DOI：10.1016/j.asoc.2024.112560.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
The computational tasks associated with Internet of Things (IoT) applications have become increasingly complex, with IoT devices (IoTDs) now being utilized in a multitude of contexts across people's daily lives. In view of the restricted computing resources and battery life of IoT devices, an increasing number of computing tasks are being handled by cloud servers. However, the restricted communication range of cloud servers gives rise to significant communication costs. Consequently, these complex computational tasks are transferred to edge servers situated in close proximity to IoT devices, with the objective of mitigating transmission delay and costs. To address the challenge of offloading decisions and obtain a multi-objective offloading decision that satisfies multiple constraints in an edge-cloud scenario, this study presents a framework of multi-objective constrained evolutionary algorithms based on membrane computing for solving constrained multi-objective computational offloading problems (MCMS-CMOEA). The framework draws inspiration from the independent parallelism observed within membrane computing membranes and the information exchange between them. Its objective is to minimize latency and energy consumption. The framework is structured into three phases, with the aim of exploring the solution space and increasing the diversity of the final solution set in a stepwise manner. In order to investigate the performance of the MCMS-CMOEA algorithm, a comparative analysis was conducted, in which the proposed algorithm was tested against nine state-of-the-art Constrained Multi-Objective Evolutionary Algorithms (CMOEAs) in three constrained multi-objective benchmark suites with different characteristics and challenges. Furthermore, the MCMS-CMOEA is compared with four offloading schemes on computational offloading problems of varying dimensions. The convergence and diversity of the algorithm are evaluated by two different comprehensive metrics: inverted generational distance (IGD) and hypervolume (HV). The numerical results of both the benchmark suite and the computational offloading problem demonstrate that the proposed algorithm exhibits superior convergence and diversity. Furthermore, the algorithm's composite results on different experiments are also more efficient and competitive, which validates the proposed algorithm as a promising approach for solving constrained multi-objective computational offloading problems.
Keywords:
Edge computing
Membrane computing (MC)
Constrained multi-objective optimization
Internet of things (IoT)
Journal
IF:
6.6
Papers:
1.4W
Citations:
4.8W