Efficient design of FIR filters using common subexpression elimination

In this paper, a novel common sub-expression elimination (CSE) method is presented for the design of linear phase finite impulse response (FIR) filters with low hardware cost. Based on weight criteria, this method eliminates the common sub-expressions of unequal weight with a transposed direct structure. By relaxing the word length while limiting its search range, the least cost coefficient sets can be generated. In order to find the optimal common sub-expressions, a matrix search process with low complexity is introduced. Compared the V-CSE method, the proposed method saves about 11.3% to 13.2% of adders on average.