Different nonlinear signal quantizations and error probability per symbol

This paper presents the analysis of different types of Laplace source signal quantizations. The original method for source quantization minimising the entropy under the mean-squared quantization error constraint is presented in the paper. The procedure for choosing constant /spl mu/ of /spl mu/ logarithmic compression law is proposed. The procedure for correct separation of the source coding from the channel one, i.e. the procedure for error probability contribution determination in the total mean-squared error (TMSE) is proposed.