New resource saving differential coefficient coding algorithm for recursive FIR filter design
Living, J and Al-Hashimi, B.M. (1999) New resource saving differential coefficient coding algorithm for recursive FIR filter design. IEEE International Symposium on Circuits and Systems, USA IEEE International Symposium on Circuits and Systems, 478-81.
Download
Full text not available from this repository.
Description/Abstract
Differential coefficient coding converts a non-recursive FIR filter into a recursive filter with the aim of reducing both maximum and average coefficient magnitudes. This paper describes a new efficient resource saving differential coefficient coding algorithm for linear phase FIR filters. The savings are obtained by not restricting the recursive section coefficients to variations of the binomial distribution as with previous approaches suitable only for narrow-band filters. Instead, an optimisation is carried out to minimise the variance of the encoded coefficients affording maximum hardware reduction for both narrow and wide-band filtering functions. Implementation of recursive FIR filters produced by the algorithm requires use of parallel filter decomposition and feedthrough accumulator structures which are both described in deatil. When compared with previous approaches, resource savings of over 40% are demonstrated.
| Item Type: | Conference or Workshop Item (UNSPECIFIED) |
|---|---|
| Additional Information: | Address: USA |
| Divisions: | Faculty of Physical and Applied Science > Electronics and Computer Science > Electronic & Software Systems |
| Item ID: | 251497 |
| Date Deposited: | 13 Apr 2000 |
| Last Modified: | 02 Mar 2012 12:57 |
| Contributors: | Living, J (Author) Al-Hashimi, B.M. (Author) |
| Date: | May 1999 |
| Additional Information: | Address: USA |
| Status: | Published |
| Publisher: | IEEE International Symposium on Circuits and Systems |
| Further Information: | Google Scholar |
| URI: | http://eprints.soton.ac.uk/id/eprint/251497 |
Actions (login required)
![]() |
View Item |


