Improved design of frequency-response-masking filters using enhanced sequential quadratic programming

Sequential quadratic programming (SQP) algorithms are widely recognized to be among the most successful algorithms for nonconvex optimization. This paper attempts to develop an SQP-based method for frequency-response-masking (FRM) filters. We explain how the complementarity conditions in the SQP algorithm help reduce the amount of computation required to update the Lagrange multipliers in a significant manner. Simulation results are presented to demonstrate the algorithm's performance that compares favorably with several existing design methods.