Learning-based multi-relay selection for cooperative networks based on compressed sensing

Abstract

Cooperative relaying is an effective technology to improve the secrecy capacity of physical-layer (PHY) security. Multiple relays can help further exploit the spatial diversity of cooperative networks. In power-limited networks, relay selection scheme is crucial important for it determines whether the optimal relay combination can be selected. This paper studies the problem of multi-relay selection in amplify-and-forward compressed sensing (AF-CS) networks, in which relays help all sources amplify and forward the signal, and the transmission matrix is used as the measurement matrix to encrypt the information. A self-organizing algorithm based on stochastic learning automata (SLA) is proposed for the AF-CS network to look for the best relay combination in a self-learning and self-optimizing way, and named “learning-based multi-relay selection algorithm” (L-MRS). In L-MRS, the destination node is self-optimizing to select the best state autonomously, and relays are self-organizing to decide whether to join the cooperation or not according to the environment feedback. Simulation studies verify the L-MRS’s is able to select the optimal relay-combination in a very stable way, and can get higher secrecy capacity compared with the coalition formation game method.