A novel implementation for narrow-band FIR digital filters

In an earlier paper Crochiere and Rabiner [1] discuss the theory of using finite impulse response (FIR) digital filters for signal decimation, interpolation, and filtering. In this paper we expand on the ideas presented in the earlier paper on implementing narrow-band designs efficiently. It is shown how, using the techniques of decimation and interpolation, a desired narrow-band filter can be realized with a greatly reduced number of multiplications per second in the realization over standard direct form implementations. Further, it is shown that the proposed implementation can have less roundoff noise and less severe coefficient sensitivity problems than a standard direct form implementation. Several examples are presented to illustrate the applicability of this implementation to practical design problems.