Linear Programming for the design of IIR filters

This paper introduces a novel procedure for the design of IIR discrete-time filters. The solution is obtained in the minimax sense through an optimization procedure termed Sequential Linear Programming (SLP). The method is based on pole-zero mapping, and the stability constraints are easily incorporated to the optimization procedure. Whereas the methodology is formulated for the design of filters that satisfy magnitude specifications, the procedure can be readily modified to suit other criteria, such as phase/group delay deviation, maximally flatness, time response, coefficient spread, or a combination of these. Efficiency of the technique is evaluated through illustrative examples and comparisons with other approaches.