Relay Selection in Wireless Networks: An Approach Inspired by the Secretary Problem

Abstract

In cooperative wireless networks with multiple relays, relay selection is an important step that has attracted a lot of research interest over the years. Many variations of the relay selection problem with different assumptions, goals, and selection approaches have been studied. In most of these approaches, it is assumed that the channel-state-information (CSI) is known. However, finding the CSI requires testing all the relays, which can be costly. This thesis proposes modeling the relay selection problem as an optimal stopping rule problem in two scenarios, one when testing to find the CSI is not costly and one where there is cost. When finding CSI has no cost, we investigate modeling the relay(s) selection problem as an optimal stopping rule problem such as a secretary problem or one of its variants. For both single relay selection (SRS) and multiple relay selection (MRS), and both known and unknown channel distributions (statistics), we show that the relay selection problem in each case can be mapped to a specific version of the secretary problem. The solution for each version differs from one another. All versions try to make the selection without testing all the channels. When testing each relay has a cost, we solve the problem using a hybrid approach. The method is based on a combination of the secretary problem and the random selection, and maximizes the overall achievable rate of the network. Simulation results verify that our method can achieve up to 7.5 dB performance gain compared to the random approach that randomly selects the relays and achieves 20 dB performance gain compared to a solution that tests all the relays and picks the best one(s).