Channel optimized predictive vector quantization

The objective of this work is to investigate how channel optimization techn iques ran be applied to predictive vector quantizers. In particular, an eff icient encoder search procedure and two design methods are derived. The des ign methods proposed here, one sample iterative and one block iterative, si multaneously optimize the predictor and the codebook, Extensive simulations show the advantage of this quantization method compared to other memory ba sed quantization schemes as well as memoryless VQ, We also demonstrate that the design methods can be used to obtain index assignments that are advant ageous to what is obtained by post process index assignment algorithms.