A Multi-Arm-Bandit Based Resource Block Allocation in RIS Assisted Wireless Network

Designing resource block (RB) allocation mechanism for reconfigurable intelligent surface (RIS) assisted wireless networks is quite limited in the preceding literature. In RI S assisted environment, a user equipment (UE) receives throughput through two channels: firstly, through the direct channel from base station (BS) to UE; secondly, through the indirect channel from BS to UE via the RIS. The existing RB allocation mechanism considers only the direct channel conditions. However, in RIS assisted environment, RB selection based on only direct channel condition may not be the optimal one. This is because, a distant UE can achieve a high throughput through the indirect channel even if the direct channel condition is not good. In this work, first, we formulate the RB allocation problem in RIS assisted environment as a multi arm bandit (MAB) problem. Then based on the MAB formulation, we develop a two-phase
$\epsilon$
-greedy algorithm which explicitly considers the throughput achievable through the indirect channel while selecting RBs. In exploitation phase, it selects the RB providing maximum reward with a probability
$1-\epsilon$
. In exploration phase, any available RB is selected randomly with a probability
$\epsilon$
. Finally, through extensive system level simulations, we have shown that our proposed algorithm performs better than an existing proportional fairness based RB allocation mechanism in terms of call dropping probability and system throughput.