A multi-agent Q-learning based rendezvous strategy for cognitive radios

摘要

In this paper, we address the blind rendezvous problem of cognitive radios (CRs) quickly finding each other to establish communication in a multi-channel dynamic spectrum access (DSA) environment. We propose a multi-agent Q-learning based rendezvous strategy that allows CR-based secondary users (SUs) to actively explore a dynamic DSA environment and learn through resulting rewards which channels are best to use for rendezvous. Through simulation, we show that our strategy enhances the rendezvous performance of SUs by enabling them to use the learned channels in an effective and efficient manner.