Analysis of a Mixed Strategy for Multiple Relay Networks

In their landmark paper Cover and El Gamal proposed different coding strategies for the relay channel with a single relay supporting a communication pair. These strategies are the decode-and-forward and compress-and-forward approach, as well as a general lower bound on the capacity of a relay network which relies on the mixed application of the previous two strategies. So far, only parts of their work - the decode-and-forward and the compress-and-forward strategy - have been applied to networks with multiple relays. This paper derives a mixed strategy for multiple relay networks using a combined approach of partial decode-and-forward with N +1 levels and the ideas of successive refinement with different side information at the receivers. After describing the protocol structure, we present the achievable rates for the discrete memoryless relay channel as well as Gaussian multiple relay networks. Using these results we compare the mixed strategy with some special cases, e. g., multilevel decode-and-forward, distributed compress-and-forward and a mixed approach where one relay node operates in decode-and-forward and the other in compress-and-forward mode.