MAP Decoding of Correlated Sources over Soft-Decision Orthogonal Multiple Access Fading Channels with Memory

We consider the joint source-channel coding (JSCC) problem where the real valued outputs of two correlated memoryless Gaussian sources are scalar quantized, bit assigned, and transmitted, without applying any error correcting code, over a multiple access channel (MAC) which consists of two orthogonal point-to- point time-correlated Rayleigh fading sub-channels with soft- decision demodulation. At the receiver side, a joint sequence maximum a posteriori (MAP) detector is used to exploit the correlation between the two sources as well as the redundancy left in the quantizer's indices, the channel's soft-decision outputs, and noise memory. The MAC's sub-channels are modeled via non-binary Markov noise discrete channels recently shown to effectively represent point-to-point fading channels. For the simple case of quantizing the sources with two levels, we establish a necessary and sufficient condition under which the joint sequence MAP decoder can be reduced to a simple instantaneous symbol-by-symbol decoder. Then, using numerical results obtained by system simulation, it is observed that when the sources are highly correlated and soft-decision quantization is used, JSCC can profit from a high correlation in the channel noise process and provide significant signal-to-distortion ratio improvements of up to 6.3 dB over a fully interleaved channel.