Adaptive filter improvement using randomly quantized coefficients

The paper is devoted to an analysis of the improvement brought to quantized LMS adaptive filters by the introduction of Random Quantization. In the first section, the deterministic quantization effects are reviewed. The second section deals with an introduction to Random Quantization and theoretical results on weight vector covariance and output noise analysis. The third section gives simulation results for a single channel equalizer application, in close agreement with the theoretical results. A controlled Random Quantization is then introduced. This makes it possible to benefit from the advantages of both deterministic and Random Quantization.