Two-Dimensional Beam Selection by Multiarmed Bandit Algorithm Based on a Quantum Walk

Abstract

This article proposes a novel beam selection method using a multiarmed bandit (MAB) algorithm based on a quantum walk (QW) principle, aimed at improving system performance. A massive multiple-input multiple-output system, employing multiple high-gain beams within a high-frequency band, is indispensable for achieving large capacity in future wireless communications. However, as the number of beams increases, selecting the most appropriate beam for each user becomes challenging due to the extensive search space and necessitating the development of a more efficient beam selection method. Therefore, we formulate a systematic process for beam selection employing the MAB algorithm rooted in QW principles. We derive the optimal parameters of this method to maximize achievable channel capacity. Through numerical analysis, we validate that the proposed method yields a greater channel capacity than that achieved not only by traditional MAB algorithms but also by an exhaustive search with overhead.