Research on Algorithms for Adaptive Antenna Arrays.

Abstract

The fundamental efficiency of adaptive algorithms is analyzed. It is found that noise in the adaptive weights increases with convergence speed. This causes loss in mean-square-error performance. Efficiency is considered from the point of view of misadjustment versus speed of convergence. A new version of the LMS algorithm based on Newton's method is analyzed and shown to make maximally efficient use of real-time input data. The performance of this algorithm is not affected by eigenvalue disparity. Practical algorithms can be devised that closely approximate Newton's method. In certain cases, the steepest descent version of LMS performs as well as Newton's method. The efficiency of adaptive algorithms with nonstationary input environments is analyzed where signals, jammers, and background noises can be of a transient and nonstationary nature. A new adaptive filtering method for broadband adaptive beamforming is described which uses both poles and zeros in the adaptive signal filtering paths from the antenna elements to the final array output.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Aug 01, 1981
Accession Number
ADA106684

Entities

People

  • Bernard Widrow
  • D. Shur
  • K. Duvall
  • R. Gooch
  • W. Newman

Organizations

  • Stanford University

Tags

Communities of Interest

  • Energy and Power Technologies
  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Adaptive Filters
  • Algorithms
  • Antenna Arrays
  • Broadband
  • Computational Science
  • Digital Filters
  • Estimators
  • Filters
  • Filtration
  • Frequency
  • Frequency Response
  • Markov Processes
  • Mathematical Analysis
  • Random Variables
  • Simulations
  • Stochastic Processes
  • White Noise

Fields of Study

  • Engineering

Readers

  • Approximation Theory.
  • Calculus or Mathematical Analysis
  • Phased Array Antenna Design.