On the design of optimal narrowband linear and minimum phase FIR filters

The design of long narrowband FIR filters is discussed in this paper. Two modifications to the popular McClellan-Parks-Rabiner linear phase filter design program are suggested. The first one is an adaptive scaling scheme in the Lagrange interpolation subroutine to overcome divergence problems. And the other change enables the design of filters with very high stopband attenuation. Because of their lower absolute delay, minimum phase FIR filters are desirable in certain telecommunications systems. The conversion of a linear phase design to minimum phase through the Herrmann-Schuessler scheme is explored. Solutions are presented to numerical stability problems encountered while using this scheme for the conversion of long filters.