Multisource Transmission for Wireless Relay Networks With Linear Complexity

This paper considers transmission schemes in multiaccess relay networks (MARNs) where J single-antenna sources send independent information to one N-antenna destination through one M-antenna relay. For complexity considerations, we propose a linear framework, where the relay linearly transforms its received signals to generate the forwarded signals without decoding and the destination uses its multiantennas to fully decouple signals from different sources before decoding, by which the decoding complexity is linear in the number of sources. To achieve a high symbol rate, we first propose a scheme called ConcurrentS→R→D-ICD in which all sources' information streams are concurrently transmitted in both the source-relay link and the relay-destination link. In this scheme, distributed space-time coding (DSTC) is applied at the relay, which satisfies the linear constraint. DSTC also allows the destination to conduct the zero-forcing interference cancellation (IC) scheme originally proposed for multiantenna systems to fully decouple signals from different sources. Our analysis shows that the symbol rate of ConcurrentS→R→D-ICD is 1/2 symbols/source/channel use and the diversity gain of the scheme is upperbounded by M-J+1. To achieve a higher diversity gain, we propose another scheme called ConcurrentR→D-ICD in which the sources time-share the source-relay link. The relay coherently combines the signals on its antennas to maximize the signal-to-noise ratio (SNR) of each source, then concurrently forwards all sources' information. The destination performs zero-forcing IC. It is shown through both analysis and simulation that when N ≥ 2J-1, ConcurrentR→D-ICD achieves the same maximum diversity gain as the full TDMA scheme in which information streams from each source are assigned to orthogonal channels in both links, but with a higher symbol rate.