Task Offloading in MEC Network by Multi-Objective Optimization Based on Load Balancing

Edge computing, a crucial infrastructure for task offloading, is gaining more attention for its low latency. However, constrained computational and storage resources often result in inequitable resource allocation. Most current studies focus on latency and energy consumption, neglecting the key metric of workload equilibrium. This paper presents a joint multi-objective optimization problem, considering time delay, energy consumption, and workload equilibrium for MEC task offloading system. To tackle this problem, we introduce an improved multi-objective grey wolf optimizer (MOGWO) combining the velocity update of particle swarm and the position update of the grey wolf, termed PSWO. Through experimental evaluations and parameter sensitivity analyses across multiple task scenarios, the proposed PSWO demonstrates superior optimization performance compared to MOEA/D-UCB, NSGAIII-SA, and MOGWO including multi-objective optimization performance and optimal single-objective function values. Furthermore, to ensure effective load balancing in computational task distribution, load variance and Jain fairness index must be the best. And, the offloading rate of edge servers ought to surpass that of end devices. The implemented system achieves an efficient offloading rate of 8 5. 9% to edge servers compared to just 1 4. 1% processed locally, thereby validating the system's effectiveness in maintaining rational and balanced resource allocation while optimizing task distribution.