Date Added: Oct 2011
In this paper, the authors use Generalized Singular Value Decomposition (GSVD) to reduce decoding complexity in an Amplify-and-Forward, Multiple-Input Multiple-Output, Two-Way Relay Channel (TWRC). The key is diagonalization of the effective channels at the source nodes; this formulation also simplifies the optimization problem of designing linear transceivers at each node to maximize the Achievable Sum Rate (ASR). For the given diagonalized structure, they first align the entries of the diagonalized channels using a permutation followed by power allocation, provided in closed form. Simulation results demonstrate that the proposed GSVD-based relaying scheme, with the signal alignment and power allocation, significantly improves the ASR while retaining the diagonalized channel structure.