A multi-agent reinforcement learning-based method for server energy efficiency optimization combining DVFS and dynamic fan control

Wenjun Lin, Weiwei Lin, Jianpeng Lin, Haocheng Zhong, Jiangtao Wang, Ligang He

Abstract: Highlights•We propose a multi-agent reinforcement learning-based method to optimize server efficiency.•A data-driven baseline comparison method is designed to improve the stability of online learning.•An improved Q-learning algorithm is proposed to address the vast state and action space.