Relay selection for improved security in cognitive relay networks with artificial noise

In this paper, we investigate the physical layer security of an underlay cognitive decode-and- forward (DF) relay network, in which one secondary transmitter (ST) exchanges confidential information with one secondary destination (SD) with the aid of multiple cognitive DF relays by sharing the licensed spectrum of one primary user (PU), and multiple colluding eavesdroppers attempt to intercept the secondary transmission. Due to deep fading, it is assumed that there is no ST-SD link, one of the cognitive relays is selected to convey the confidential information, while the others emit artificial noise (AN) in an effort to confuse the eavesdroppers. In the absence of the eavesdroppers' channel information state (CSI), we analyze the secrecy outage performance under two opportunistic relay selection schemes, which are the conventional relay selection (CRS) and the best relay selection (BRS) schemes. We derive novel closed-form expressions for the secrecy outage probability with these two relay selection schemes. Simulation results are provided to verify our analysis, and shows that the secrecy performance of the propose BRS scheme significantly outperforms the CRS one.