| This work studies several relay network geometries whose opportunistic diversity-multiplexing tradeoff (DMT) has been unknown. In the past, opportunistic analysis has been applied to multi-user diversity and relay selection based on methods of order statistics. These techniques rely on assumptions that break down in other interesting and useful network topologies, for example the opportunistic user selection in a n × n network with a relay (among many others). This work presents techniques that expand opportunistic analysis to a wider set of networks. New opportunistic methods are proposed for several network geometries and analyzed in the high signal noise ration (SNR) regime. For each of the relay geometries in the work, we study the opportunistic DMT of several relaying protocols, including amplify-and-forward, decode-and-forward, compress-and-forward, non-orthogonal amplify-forward, and dynamic decode-forward. Among the highlights of the results: in a variety of multi-user single-relay networks, simple user election strategies are developed and shown to be DMT-optimal. It is shown that compress-forward relaying achieves the DMT upper bound for the opportunistic multiple-access relay channel as well as in the n × n user network with relay. Other protocols, e.g. dynamic decode-forward, are shown to be near optimal in several cases. Finite- precision feedback is analyzed for the opportunistic multiple-access relay channel, the opportunistic broadcast relay channel, and the opportunistic gateway channel, and is shown to be almost as good as full channel state information. In heterogeneous relay channels, it is shown that mixing relays with simple protocols, like amplify and forward, with relays with more complicated protocols, like dynamic decode and forward or compress and forward, can help achieve gains from both types of relays. For the bi-directional multi-relay channel with a direct link between the two sources, we propose a dynamic-decode and forward opportunistic scheme and prove that it is optimal at high multiplexing gains. |
