Worst-Case Linear Transmit Strategies with Limited Relay Information in Gaussian Relay Networks

In practical large scale relay networks, it is often an unfeasible task to adapt the strategy of the relays to each source/sink pair communicating through the network. In this paper we consider the strategy of the relays as fixed and chosen in advance. In this context, it is a crucial problem for the source node transmitting information, to know the strategy of the relays in the network. Therefore we optimize the transmit strategy of the source node, assuming only partial relay information knowledge, i.e., channel state information and relay strategies. We also study the impact of getting complete relay information from certain relays only, derive the reduction in the problem uncertainty due to this knowledge and evaluate numerically the performance of the network.