An Asymptotically Capacity-Achieving Scheme for the Gaussian Multiple-Access Relay Channel

We study the multiple access relay channel (MARC) with relay-destination cooperation (MARC-RDC). This channel resembles the uplink transmission in heterogeneous network where two user equipments (UEs) communicate with macro-cell base station (BS) through small-cell BS. We propose a coding scheme where the transmission is carried over B blocks and each UE performs superposition block Markov encoding. The relay (small-cell BS) first jointly decodes both UEs information using sliding window decoding over two transmission blocks and then forwards these information to the destination (macro-cell BS) coherently with UEs. The destination quantizes its received signal in each block and forwards the quantization index to the relay. The destination then decodes both UEs information using backward decoding. For this scheme, we derive the achievable rate region and compare it with existing schemes and the cut-set bound. Results show that relay-destination cooperation enlarges the rate region as the destination power increases. We further show that the proposed scheme asymptotically achieves the capacity by reaching the cut-set bound when the destination power approaches infinity and the ratio of one UE-destination to UE-relay link amplitudes is equal to that of the other UE. These results make the proposed scheme appealing for deployment in 5G cellular networks.