Source Side Information Can Increase the Joint Source-Channel Coding Error Exponent

Abstract

Abstract-We study the joint source-channel coding (JSCC) error exponent for discrete memoryless source-channel systems with side information which is correlated to the transmitted source. Two cases are considered: (1) the side information is available only at the decoder; (2) the side information is available at both the encoder and decoder. We employ the method of types to establish a lower bound for the JSCC error exponent for each case. As a consequence, a JSCC theorem on the reliable transmissibility of the source over the channel is obtained. It is noted that the same JSCC theorem applies for both cases. For binary sources and symmetric channels, we derive a sufficient condition for which the side information at the decoder can strictly improve the JSCC error exponent. Numerical results show that side information can enlarge the region for reliable transmissibility and increase the JSCC error exponent for a wide class of source-channel parameters.

Joint source-channel with side information coding error exponents

“…The more general case, where W Y |X is a noisy channel, is recently studied [15], [14]. It is shown that, not surprisingly, the reliable reconstruction of a n is possible if and only if the channel capacity of the channel is larger than the conditional entropy of the source.…”