Interference Truncation for Compute-and-Forward Relaying in Multiple Access Channels

Nested lattice code is a practical technique to approach the Shannon capacity upper bound. However, the performance of compute-and-forward (CF) relaying is blocked by the channel output residuals that the decoding process of nested lattice codeword cannot remove completely. In this paper, an interference truncation scheme is designed for CF relaying in multiple access channels to minimize the channel residual caused by lattice decoding, where the scalars and coefficient vectors are configured to optimize the computation rate. The key idea is to improve the performance by introducing a recovered linear combination of codewords into the decoding process at the destination with proper scaling. The proposal harnesses the received message to reduce the impact of effective noise, which in traditional CF is determined by that equations are computed out directly from received messages. The analysis and numerical results demonstrate that the achievable computation rate of the proposed scheme outperforms the traditional scheme, and the impacts of asymmetric transmit power settings are studied as well.