Multiple-access relay channels with non-causal channel state information at the relay

In this paper, we study state-dependent multiple-access relay channel (MARC). (i) Inner and outer bounds are derived for two-user state-dependent discrete memoryless MARC in which channel state information is known non-causally at the relay (informed relay). The inner bound is obtained by using decode and forward strategy and a combination of binning scheme and codeword splitting. Also, by applying Fano's inequality, we obtain an outer bound for state-dependent MARC with channel state at the relay. (ii) Furthermore, we extend the results of discrete alphabet to Gaussian case as a discrete time and continuous alphabet model. Our general Gaussian results subsume previous special ones and specifically gives a general version for the Costa's theorem. (iii) We show numerically that the achievable rate region of Gaussian MARC with cognition of the relay about interference is larger than the achievable rate region of Gaussian MARC without any knowledge about interference. Also, by numerical examples, we investigate the capacity-achieving case in high signal to noise ratios at the relay. Copyright (c) 2014 John Wiley & Sons, Ltd.