Adaptive UCB MAB Algorithm forOptimizing Relay Selection in Narrowband Power Line Communication

In this paper, we aim to apply artificial intelligence in order to solve the relay selection problem for two-hop cooperative narrowband Power Line Communication (NB-PLC), without channel side information. The communication channel is corrupted by periodic impulsive noise, which is modeled using Middleton Class-A noise. The main idea is to enhance the conventional multi-armed bandits (MAB) algorithm, called Upper-Bound Confidence (UCB), by integrating a classification process. This modification is intended to enable the UCB algorithm to adapt to non-stationary situations and address its limitations, specifically the requirement for multiple sequential tests to determine the optimal value of the main parameter in UCB which is instrumental in controlling the exploration level. The simulation results demonstrate that the proposed new algorithm outperforms the conventional UCB algorithm in terms of cumulative regret, probability of good selection, and Bit Error Rate (BER).