Index assignment for MDVQ over memoryless binary symmetric channel with packet erasure

Abstract

Genetic algorithm is applied to index assignment for multiple description vector quantization (MDVQ) over memoryless binary symmetric channel (BSC) with packet erasure. The cost evaluation function is presented based on this hybrid model. Simulation results based on the bit error rate for binary symmetric channel and the packet erasure rate demonstrate the usefulness of the proposed optimization scheme. © IEICE 2006.