$n$ -Channel Asymmetric Entropy-Constrained Multiple-Description Lattice Vector Quantization

摘要: This paper is about the design and analysis of an index-assignment (IA)-based multiple-description coding scheme for n-channel asymmetric case. We use entropy constrained lattice vector quantization restrict attention to simple reconstruction functions, which are given by inverse IA function when all descriptions received or otherwise a weighted average descriptions. consider smooth sources with finite differential rate MSE fidelity criterion. As in previous designs, our construction based on nested lattices combined through single function. The results exact under high-resolution conditions asymptotically as nesting ratios approach infinity. For any n, optimal within class IA-based schemes. Moreover, case two dimensions greater than one, performance strictly better that existing designs. In three descriptions, we show limit large dimensions, points inner bound Pradhan can be achieved. Furthermore, yields, symmetric case, smaller loss recently proposed source-splitting approach.