DIGITAL SYNTHESIS OF NONLINEAR FILTERS,

Abstract

The paper presents a technique for constructing the optimal nonlinear filter using a digital computer. The viewpoint differs fundamentally from almost all previous work in that the state of the nonlinear estimator, the conditional density of the signal given the observations, is represented as a set of point masses on a moving grid. Although the method leads to very accurate realizations of optimal nonlinear filters, there is an inherent drawback in that on exponential dependency of computation time upon dimension rapidly causes the proposed method to become computationally intentable as the dimension increases. This paper describes the methods in detail and indicate how they apply to a particular problem of a passive receiver design. For the particular problem, a Monte Carlo error analysis shows that the optimal filter achieves an order of magnitude mean-square error improvement over the extended Kalman-Bucy filter and at least two orders of magnitude improvement over the least squares filter. Finally a discussion of the implications of these studies on the design of digital machines for filtering purposes is presented. (Author)

Document Details

Document Type
Technical Report
Publication Date
Jul 01, 1970
Accession Number
AD0708983

Entities

People

  • K. D. Senne
  • R. S. Bucy

Organizations

  • University of Southern California

Tags

DTIC Thesaurus Topics

  • Algorithms
  • Computations
  • Computers
  • Digital Computers
  • Error Analysis
  • Errors
  • Estimators
  • Filters
  • Filtration
  • Mathematical Analysis
  • Mathematics
  • Observation
  • Optimal Estimators
  • Statistical Algorithms

Fields of Study

  • Engineering

Readers

  • Adaptive Control and Estimation with Uncertainty in Dynamic Systems.
  • Finite Element Method (FEM) for solving Partial Differential Equations (PDEs)