Sum Capacity Scaling Law of Opportunistic Relaying for MIMO Wireless Communication

We consider a setup where a source communicates with a destination, both equipped with M antennas, assisted by K single-antenna relays using an amplify-and-forward half-duplex protocol. In contrast to conventional approaches where all relays in the network participate, we propose a new opportunistic relaying protocol in which only a few very important relays (VIRs) are selected to share the total power. Assuming perfect channel state information (CSI) at the destination and CSI of the backward channels for the selected VIRs at the source node, we show that the network capacity scales as M/2 log log K + O(1) for fixed M, fixed signal-to-noise ratio, and K → ∞. We also propose two low complexity algorithms for selecting the VIRs, and show that our proposed opportunistic relaying protocol can achieve the optimal capacity scaling, and also yield significant capacity gains compared with the conventional approach.