Affine index assignments for binary lattice quantization with channel noise
A. Mehes, K. Zeger · 2002
Two major issues in noisy channel vector quantization are complexity and sensitivity to channel errors. Structured vector quantizers and index assignments provide a low complexity solution for enhancing channel robustness. A general formula is given for the MSE performance of affine index assignments for a binary symmetric channel with an arbitrary source and a binary lattice quantizer. The result is then used to compare some well-known redundancy free codes. The binary asymmetric channel is considered for a uniform input distribution and a class of affine codes.