A fast algorithm for FIR digital filtering with a sum-of-triangles weighting function (Q1911852)

A new algorithm for designing digital finit impulse response (DFIR) filters is presented. It is known that apart some significant advantages, DFIR filters lead to a relatively high volume of computation. There are now some methods to overcome this difficulty, such as the fast Fourier transform (FFT) approach and the direct convolution one using dedicated hardware. The author proposed a new method whose underlying idea is to approximate the desired weighting function (WF) by a sum of triangle-like WFs. It is shown that by this way, one can associate a stable recursion and at the same time very few multiplications are required. Some numerical examples show the good capability of such an approximation. The method is undoubtly very interesting but the way to choose the triangle is not very clear.