Dynamic Control of Functional Splits for Energy Harvesting Virtual Small Cells: a Distributed Reinforcement Learning Approach

TL;DR

This paper introduces a distributed reinforcement learning approach to dynamically control functional splits in energy-harvesting virtual small cells, optimizing energy use and system performance in a network with edge computing coordination.

Contribution

It develops multi-agent RL algorithms for energy-efficient control of virtual small cells with energy harvesting, demonstrating improved coordination and adaptation over uncoordinated methods.

Findings

Coordination via broadcasting improves system-level gains.

Distributed RL achieves rewards close to offline bounds.

Continuous state/action representation enhances learning speed and adaptability.

Abstract

In this paper, we propose a network scenario where the baseband processes of the virtual small cells powered solely by energy harvesters and batteries can be opportunistically executed in a grid-connected edge computing server, co-located at the macro base station site. We state the corresponding energy minimization problem and propose multi-agent Reinforcement Learning (RL) to solve it. Distributed Fuzzy Q-Learning and Q-Learning on-line algorithms are tailored for our purposes. Coordination among the multiple agents is achieved by broadcasting system level information to the independent learners. The evaluation of the network performance confirms that coordination via broadcasting may achieve higher system level gains than un-coordinated solutions and cumulative rewards closer to the off-line bounds. Finally, our analysis permits to evaluate the benefits of continuous state/action representation for the learning algorithms in terms of faster convergence, higher cumulative reward and more adaptation to changing environments.