On FIR digital filter implementations for interpolation

In this correspondence we examine some problems on the implementation of digital interpolator using linear-phase FIR filters. A procedure for selecting design parameters such as sampling rate increase ratio L, filter impulse response length N is presented. It is shown that, except for one interpolated sample in the case where L is even and (N+1)/L, is odd, all (L - 1) interpolated samples can be computed from the same set of original input samples. This fact can simplify the design of the control section of interpolation filters. The symmetry property of the impulse response of linear-phase FIR filters is exploited. It is shown that, by suitably arranging arithmetic operations, the number of multiplication operations can be reduced about 25 percent.