Digital Signal Processing Reference
In-Depth Information
Chapter 7
Iterative Design of FIR Filters
Bipul Luitel and Ganesh Kumar Venayagamoorthy
Abstract
This chapter presents the iterative design of finite impulse response (FIR)
filters using particle swarm optimization with a quantum infusion (PSO-QI) algo-
rithm. Filter design, in this work, is formulated as a parameter optimization problem
using population-based stochastic methods; and hence, it is iterative. PSO-QI is a
hybrid algorithm combining PSO and quantum-behaved PSO. PSO-QI combines
the best features of these individual algorithms. Therefore, the design specification
for FIR filters can be satisfied more accurately. Two methods of evaluating the per-
formance of the algorithm (cost function) are implemented. Minimizing the mean
squared error between the actual and the ideal filter response is one approach to per-
formance evaluation. The second approach involves minimizing the mean squared
error between the ripples in the passband and the stopband of the designed filter
and the desired filter specification. The results presented show that filters designed
using PSO-QI most closely match the design specification, and their performance is
more consistent when compared with other evolutionary algorithms. The results are
compared with the constrained least squares method of filter design.
7.1 Introduction
Digital filters suppress the unwanted parts of a signal, such as noise, and extract the
important parts, such as underlying components within a frequency range. Conse-
quently, they have been applied in communication for noise reduction, audio/video
signal enhancement, etc. Digital filters are an important component of digital sig-
nal processing and are used in a wide range of modern applications including, but
not limited to, telecommunications, acoustics, biometrics, biomedical science, and
speech and image processing.
Traditionally, techniques such as the tables method [
7
], windowing method [
1
],
frequency sampling method, and best uniform approximation method have been
B
B. Luitel (
G.K. Venayagamoorthy
Real-Time Power and Intelligent Systems (RTPIS) Lab, Clemson University, Clemson, SC, USA
e-mail:
iambipul@ieee.org
G.K. Venayagamoorthy
e-mail:
gkumar@ieee.org
)
ยท
Search WWH ::
Custom Search