Accession Number : AD0753119

Title :   Optimum Constrained Order Digital Filters.

Descriptive Note : Scientific rept.,

Corporate Author : NORTH CAROLINA STATE UNIV RALEIGH DEPT OF ELECTRICAL ENGINEERING

Personal Author(s) : Cheng,Stephen Miau

Report Date : 10 NOV 1972

Pagination or Media Count : 60

Abstract : In practical applications of digital filters, the order of the filter is an important consideration because it drastically affects the cost, complexity, speed and accuracy of the filter. In filters designed to perform minimum mse estimation, the tradeoff between order and mse is often difficult. This dissertation investigates filter design algorithms for constrained order but nearly optimum estimation. It covers the problems of interpolation, smoothing and extrapolation of a process given noise measurements of that process. The design algorithm is derived and a method of implementation is suggested. The properties of the filter are examined. The filter is applied to recursive prediction in differential pulse code modulation and delta modulation communication systems. It is also used to design pre-emphasis and de-emphasis networks in a pulse code modulation system. A most interesting application is the estimation of power spectra using the constrained order filter as a whitening filter. (Author)

Descriptors :   (*INFORMATION THEORY, TIME SERIES ANALYSIS), PULSE CODE MODULATION, INTEGRAL TRANSFORMS, TRANSFER FUNCTIONS, DIGITAL SYSTEMS, DIFFERENCE EQUATIONS, POWER SPECTRA, WHITE NOISE, INTERPOLATION, STOCHASTIC PROCESSES, CURVE FITTING, GRAPHICS, COMPUTER PROGRAMS

Subject Categories : Cybernetics

Distribution Statement : APPROVED FOR PUBLIC RELEASE