Accession Number : AD0488877

Title :   REPRESENTATION AND ANALYSIS OF SIGNALS. PART XXII. TECHNIQUES FOR THE RESOLUTION OF OVERLAPPING WAVEFORMS.

Descriptive Note : Technical rept.,

Corporate Author : JOHNS HOPKINS UNIV BALTIMORE MD DEPT OF ELECTRICAL ENGINEERING AND COMPUTER S CIENCE

Personal Author(s) : Reeder, James R.

Report Date : MAY 1966

Pagination or Media Count : 187

Abstract : Many signals arising from physical processes are composed of overlapping waveforms. This dissertation provides a unified approach to the problem of resolving the individual waveforms comprising such signals. The techniques presented are classified as (1) statistical parameter estimation techniques, (2) Fourier techniques, (3) techniques for resolving truncated waveforms of generalized exponential type. Practical computational algorithms are presented for implementing the parameter estimation techniques using the method of nonlinear least-squares estimation. An analogy is shown to exist between the resolution problem and the singular detection problem of statistical communication theory. A filter is derived from the Fourier techniques which permits perfect resolution of the waveforms in the singular case and significant resolution enhancement in the nonsingular case. A test statistic is derived from the principle of analytic continuation and the method of maximum likelihood testing. An approximate resolution technique is developed for a special class of ESR signals which possesses significant computational advantages over some standard methods of analysis.

Descriptors :   *SIGNALS, SEISMOLOGY, NUCLEAR MAGNETIC RESONANCE, ELECTRONS, PARAMAGNETIC RESONANCE, INFRARED SPECTROSCOPY, MICROWAVE SPECTROSCOPY, ELECTROPHORESIS, MULTIPATH TRANSMISSION, RESOLUTION, OPTIMIZATION, MATHEMATICAL MODELS, INTEGRAL TRANSFORMS, ALGORITHMS, LEAST SQUARES METHOD, NOISE, DETECTION, SEPARATION, FOURIER ANALYSIS.

Subject Categories : Acoustic Detection and Detectors

Distribution Statement : APPROVED FOR PUBLIC RELEASE