An Efficient Strategy for Applying Compute-and-Forward to the MARC

With the aim of improving network throughput and achieving full diversity gain, this paper focuses on strategies for applying compute-and-forward (CF) scheme to the multiple access relay channel (MARC). The direct application of the original CF to the MARC results in poor error performance due to the probability of a rank deficient coefficient matrix, which is an inherent problem of a CF system. One way to solve this problem is by allowing the relay and the destination to fully cooperate with each other in constructing a full rank coefficient matrix. However, this requires a large amount of communication overhead. This paper proposes an efficient strategy where the destination always attempts to decode transmitted messages by itself, without the help of the relay. The destination cooperates with the relay only when necessary. It is shown that with a small amount of overhead, the proposed strategy outperforms the existing approaches in terms of achievable sum-rate, outage probability, and throughput. Furthermore, the proposed strategy also achieves the full diversity gain of the MARC.