Game-based resource allocation in heterogeneous downlink CR-NOMA network without subchannel sharing

This paper explores how to manage resources in a two-tier, diverse cognitive radio network using a sophisticated technique called non-orthogonal multiple access. In this setup, secondary base stations (BSs), equipped with cognitive capabilities, act as bridges between the main BS and the secondary users (SUs). The main goal is to maximize the throughput of the secondary network by efficiently relaying information to capable SUs without the knowledge of their channel characteristics. To achieve this, we set out to solve a complex problem involving maximizing throughput while staying within transmission power limits, meeting minimum data rate requirements, and controlling interference power. Our approach introduces a new SU clustering method based on fuzzy logic, paired with a hierarchical learning method inspired by the well-known UCB1 algorithm to allocate power effectively to these clusters. We frame this joint optimization of SU cluster formation and power allocation as a cooperative multi-armed bandit game. Through extensive simulations, we demonstrate the effectiveness of our proposed methods. We also explore how different parameters affect the performance of the system, shedding light on the intricate workings of this network architecture.