Multi-User Position Based on Trajectories-Aware Handover Strategy for Base Station Selection with Multi-Agent Learning

This paper presents the optimal Base Station(BS) selection method for proactive decision handover(HO) in Millimeter-wave(mm-wave) wireless communication. Mm-wave spectrum suffers significantly from the high path-loss and blockage caused by either controlled or uncontrolled sources. While the primary purpose of utilizing mm-wave is to achieve a high data rate, the presence of obstacle degrade the overall system performance since the connection link between User(UE) and serving BS being intermittent. The repercussion of the sporadic link is an increased number of HO. To increase throughput, proactive HO and minimize unnecessary HO are considered as the solution, and this paper presents a solution based on Reinforcement Learning(RL) framework. The framework learns from multi UE trajectories, and smart-agent learns simultaneously using Multi-Agent RL(MARL) and mapping each trajectory's feature and respect Q-value in smart agent constructed from Artificial Neural Network(ANN). The numerical results show that the intelligent, learned agent minimizes the number of HO and also outperform heuristic HO strategy in terms of throughput.