Minimizing the Worst-Case Distortion in Channel Splitting

situations are possible, but the encoder does not know which of these K situations is realized. The decoder uses the streams from the operative channels to form a sequence of reconstructions X in a measurable space SC, (often, but not necessarily, the same as 3C). Performance is measured by the time average of a distortion function d(X , Xt). For any given coding scheme, the expected distortion will depend on the t t *-£(") 1979 (2)