Improving the Manry-Aggarwal Method for Designing Multi-Dimensional FIR Digital Filters,

Abstract

M. T. Manry and J. K. Aggarwal recently described an algorithms for use in the design of multi-dimensional FIR digital filters by phase correction. As they observe, their method can be viewed as the steepest descent method for minimizing a certain function f(x): given an approximate solution x(n), a new approaximation is x(n + 1) =x(n) + t(n)p(n) where p(n) = grad f(x(n)) and t(n) is chosen by a simple rule. We derive here an improved rule for determining t(n) and an improved direction p(n) (essentially the Fletcher-Reeves conjugate-gradient direction). The resulting method appears to be two to three times as fast as the Manry-Aggarwal method; the additional cost is primarily in storage which roughly doubles. (Author)

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Sep 01, 1976
Accession Number
ADA034541

Entities

People

  • James W. Daniel

Organizations

  • University of Texas at Austin

Tags

DTIC Thesaurus Topics

  • Algorithms
  • Amplitude
  • Coefficients
  • Computations
  • Digital Filters
  • Discrete Fourier Transforms
  • Equations
  • Filters
  • Frequency Domain
  • Iterations
  • Mathematics
  • Military Research
  • Numbers
  • Numerical Analysis
  • Square Roots
  • Steepest Descent Method
  • United States

Readers

  • Analytical Mechanics
  • Approximation Theory.
  • Operations Research