Performance of switching-based partial relay selection scheme for amplify-and-forward cognitive relay networks

The outage probabilities (OPs) of cognitive amplify-and-forward systems with the conventional and the switching-based partial relay selection (PRS) schemes over independent but not identically distributed Rayleigh fading channels are evaluated. The conventional PRS scheme always depends on the instantaneous channel quality of the first hop, while the switching-based PRS scheme is based on the estimated average channel state information (CSI) of the first and second hops. For the switching-based PRS scheme, the instantaneous CSI for links with the smaller average channel power in the first and the second hops for each end-to-end path is used for relay selection. Thus, the switching-based PRS scheme counts on the instantaneous CSI of either the first or second hop. The tight lower bounds and asymptotic expressions of the OP are derived. The feedback overheads of both schemes are discussed. Simulation results substantiate the theoretical analysis and also reveal that the switching-based PRS outperforms the conventional one over the cognitive relay networks in terms of OP.