Ultrafast Resource Allocation by Parallel Bandit Architecture Using Chaotic Lasers for Downlink NOMA Systems

Abstract

The effectiveness of laser chaos decision-makers in facilitating ultrafast decision-making makes it possible for a real-time process of channel allocation scheme for the non-orthogonal multiple access (NOMA) technology in the next generation of wireless communications. However, managing the increasing number of users is challenging as the complexity of channel allocation increases significantly. To resolve this challenge, this paper proposes a novel approach to address scalability problems by introducing a parallel bandit architecture using an array of laser chaos decision-makers. In the proposed method, each user is allocated a specific channel by a dedicated laser chaos decision-maker, thereby reducing the number of available options compared with the conventional approach. This parallel bandit architecture enables the system to efficiently manage increasing users while maintaining scalability and ultrafast channel allocation in NOMA. Additionally, fairness is considered by incorporating a logarithmic utility function for design compensation. Numerical simulation results demonstrate that the proposed method achieves higher throughput and enhanced fairness than conventional NOMA approaches such as minimum distance-NOMA (MD-NOMA), conventional-NOMA (C-NOMA), and uniformed channel gain difference-NOMA (UCGD-NOMA). Moreover, the system performance is evaluated on a larger scale, accommodating a significant number of users, with the study considering up to 64 users, surpassing the limitations of the conventional approach of one laser chaos decision-maker, which is constrained to 10 users. INDEX TERMSNon-orthogonal multiple access (NOMA), channel allocation, logarithmic utility function, laser chaos decision-maker, multi-armed bandit problem I. INTRODUCTION VOLUME XX, XXXX