Showing papers by "Martin Morf published in 1983"

Estimation of time differences of arrival by pole decomposition

Abstract: Time differences of arrival (TDOA's) of emitter wave fronts to a spatially distributed array of sensors can be used to determine the source location. In this paper, we suggest a new method of TDOA estimation for multiple unknown autoregressive moving average (ARMA) sources and additive noise that may be correlated between the sensors. We derive a theoretical formula that only uses the receiver cross spectra and the source poles for the TDOA determination. The poles are estimated by a least squares technique, and two methods are suggested for the estimation of the cross spectra which allow tradeoffs between computational complexity and accuracy. A new time delay model is derived and used to show the applicability of the methods for noninteger TDOA's. Results from simulations illustrate the performance of the algorithm by Monte Carlo trials and compare it to the Cramer-Rao bound.

A relationship between the Levinson algorithm and the conjugate direction method

Abstract: This note provides a connection between the (block) Levinson algorithm and the conjugate direction method (CDM). We first present an extension of the scalar CDM to the block matrix case. Then we show how the (block) Levinson recursions can be derived as a special case of the (extended) CDM. This gives a new derivation and interpretation for Levinson's algorithm.

A new derivation for fast recursive least squares and Levinson algorithms by the conjugate direction method

Abstract: This paper provides a new derivation and interpretation for the fast recursive least squares (RLS) algorithm of [1-3] and for the (block) Levinson algorithm as specialized conjugate direction methods (CDMs). The fast RLS time update is shown to be equivalent to a particular combination of two last steps in CDM recursions, giving a novel geometric description for this method. The results for the Levinson algorithm extend the ones of [9] on its relationship to the CDM.

A tree architecture for the symmetric eigenproblem

Abstract: In this paper we present a tree architecture foreigendecomposition based on a Divide -and-Conquer approach. A comparison of this method with the previously proposed systolic array architecture is made. Strategies for serial -parallel computation and fault -tolerantdesign are indicated. The suitability of this architecture for VLSI implementation is pointed out. 1. Introduction 1.1. Problem Description Many signal and image processing algo-rithms require the eigendecomposition of large matrices for efficient application [1,2,3]. Often, these problems have to be solved in a distri- buted fashion, e.g. in a Distributed Sensor Net- work involving the combination of sub -spaces. Another notable example is the so- called 'super- radar'. There have been many solutions to the eigendecomposition problem. We will be concerned mainly with the symmetric eigen- problem. The quantities of interest are the eigenvalues (all or some of them) and the corresponding eigenvectors. The specializationto symmetric matrices is not overly restrictive,

Concurrent Implementations Of Matrix Eigenvalue Decomposition Based On Isospectral Flows

Abstract: In this paper we evaluate several techniques for solving the symmetric tridiagonal problem based on the method of isospectral flow. Architectures which result from these considerations are discussed. Their advantages and disadvantages from the viewpoints of numerical accuracy and ease of implementation in VLSI are also investigated.© (1983) COPYRIGHT SPIE--The International Society for Optical Engineering. Downloading of the abstract is permitted for personal use only.

Recursive identification algorithms for right matrix fraction description models

Abstract: Multivariable identification algorithms are usually designed only for left matrix fraction description (LMFD) models. In this paper we consider recursive identification algorithms for right matrix fraction description (RMFD) models with diagonal denominator matrices. The algorithms are of prediction error (PE) and model reference (MR) type. Convergence analysis results give a positive realness condition for the convergence of the MR algorithm to the true RMFD of the system. The PE scheme converges to a local minimum of the criterion with probability 1. Results from simulations illustrate the convergence of the algorithms.

Development of signal processing algorithms via functional programming

Abstract: Motivated by recent advances made in VLSI and VLSI-based systems, we describe the functional programming(FP) style as proposed by Backus [1]. Comparative studies are made by coding simple algorithms in a conventional language and in FP. We find that the methodologies involved in the two approaches are different; conventional languages require thinking at an object level while FP styles favour a more hierarchical formalization. The paper also discusses some limitations of FP styles that will have to be remedied.