Optimal Task Allocation Using Manta Ray Foraging And Genetic Algorithms In Heterogeneous Edge Environments

Abstract

Edge computing has evolved as a solution for latency-sensitive IoT applications, yet efficient task allocation is challenging due to dynamic workloads, heterogeneous resources, and fluctuating network conditions. This paper proposes a novel integration of Manta Ray Foraging Optimization with Genetic Algorithm (MRFO-GA) for optimal task allocation. The fuzzy membership functions and rules prioritize tasks based on CPU utilization, storage utilization, bandwidth availability, network latency, and deadline constraints. We employ Markov Decision Process (MDP) and Lyapunov drift theory to achieve effective offloading decisions and load balancing, respectively. We compare the proposed MRFO-GA approach against four state-of-the-art approaches using five evaluation metrics: Energy (J), Latency (ms), Fitness Score, Packet Loss (%), and Reliability (%). The proposed MRFO-GA demonstrates superior performance across task loads of 50, 300, and 450.