CONSTRAINED LEAST SQUARES DESIGN OF M-D FIR FILTER BASED ON CONVEX PROJECTION TECHNIQUES

概要

論文の詳細を見る
In this paper, we propose a class of novel design methods of "constrained least squared M-D FIR filter". The most remarkable feature is that all of the proposed methods have mathematical guarantees of convergences to the unique least squared error solution under any consistent set of prescribed maximum error condition. The proposed methods are based on "convex projection techniques" that compute the metric projection onto the intersection of multiple closed convex sets in a real Hilbert space. Some of the proposed methods can still be applied to a design problem on which any inconsistent set of maximum error conditions are imposed. In such a case, these methods lead to the unique least squared error solution over the set of all filters that attain the least sum of squared distances to all constraint sets.
社団法人電子情報通信学会の論文
1998-01-20