Game theory-based virtual machine migration for energy sustainability in cloud data centers

As the demand for cloud computing services increases, optimizing resource allocation and energy consumption has become a key factor in achieving sustainability in cloud environments. This paper presents a novel approach to address these challenges through an optimized virtual machine (VM) migration strategy that employs a game-theoretic approach based on particle swarm optimization (PSO) (PSO-GTA). The proposed approach leverages the collaborative and competitive dynamics of Game Theory to minimize energy consumption while using renewable energy. In this context, the game is represented by the swarm, where each player, embodied by particles, carries both competitive and cooperative elements essential to shape the collective behavior of the swarm. PSO is integrated to refine migration decisions, improving global convergence and optimizing the allocation of VMs to hosts. Through extensive simulations and performance evaluations, the proposed approach demonstrates significant improvements in resource utilization and energy efficiency, promoting sustainability in cloud computing environments. This research contributes to the development of environmentally friendly cloud computing systems, thus ensuring the delivery of energy-efficient cloud computing. The results demonstrate that the proposed approach outperforms fuzzy and genetic methods in terms of renewable energy usage. The PSO-GTA algorithm consistently outperforms Q-Learning, Pittsburgh and KASIA across three simulation scenarios with varying cloudlet dynamics, showcasing its efficiency and adaptability, and yielding improvements ranging from 0.68% to 5.32% over baseline results in nine simulations.