Weighted least-squares implementation of Cohen-Posch time frequency distributions with specified conditional and joint moment constraints

Abstract

A positivity constrained iterative weighted least-squares (WLS) method for constructing nonnegative joint time-frequency distributions (i.e., Cohen-Posch TFDs) satisfying marginal, joint moment, conditional moment and generalized marginal constraints is developed. The new algorithm solves the leakage problem of the least-squares approach and is computationally faster. It is also more computationally efficient than the MCE implementation of these constraints, developed by Loughlin, Pitton and Atlas (1994).