Strategies for source/sink pairs in amplify-and-forward Gaussian relay networks

This paper presents a framework for studying amplify-and-forward Gaussian relay networks and optimizing the strategy of source/sink pairs. A typical communication scenario is that many pairs of users try to exchange information over the same network. Changing the strategy of the relays, in such a network, for each communication might be an overwhelming task. We consider the case where the behavior of the relays is fixed. By solely optimizing the strategy of the source and the sink, the goal is to maximize the mutual information between the transmitted signal and the received signal. We show a parallel between the multiple-input multiple-output (MIMO) single user channel and amplify-and-forward Gaussian relay networks and derive solutions for three different types of power constraints on the network, namely 1) source power constraint, 2) global network power constraint and 3) individual relay power constraints. Finally we provide numerical results for an example network.