An Iterative Constrained Minimax Approach to Magnitude Response Design of FIR Evidence Filters

摘要

Evidence filtering is a promising approach to infer the “frequency” characteristics of various events of interest from temporally and spatially distributed sensor data based on Dempster–Shafer evidence theory. The design of evidence filters has several challenges due to the nonnegativity of the filters’ coefficients. This paper presents an iterative constrained minimax method for the magnitude response design of finite impulse response evidence filters with a prescribed transition-band magnitude drop. The method iteratively converts the problem into a sequence of minimax magnitude error subproblems of the evidence filters with given minimum stopband attenuations. The convergence of the global solutions of these subproblems to the global solution of the original problem is established. By using a recently published algorithm to solve the subproblems, the proposed method has obtained better magnitude responses than existing methods in design examples of this paper.