We consider zero-delay single-user and multi-user source coding with averagedistortion constraint and decoder side information. The zero-delay constrainttranslates into causal (sequential) encoder and decoder pairs as well as theuse of instantaneous codes. For the single-user setting, we show that optimalperformance is attained by time sharing at most two scalar encoder-decoderpairs, that use zero-error side information codes. Side information lookaheadis shown to useless in this setting. We show that the restriction to causalencoding functions is the one that causes the performance degradation, comparedto unrestricted systems, and not the sequential decoders or instantaneouscodes. Furthermore, we show that even without delay constraints, if either theencoder or decoder are restricted a-priori to be scalar, the performance losscannot be compensated by the other component, which can be scalar as wellwithout further loss. Finally, we show that the multi-terminal source codingproblem can be solved in the zero-delay regime and the rate-distortion regionis given.
展开▼