Algorithms and Structures for Real-Time Signal Processing. Revised.

摘要

The study of digital signal processing is to a large extent the study of algorithms, but there is a natural classification into algorithms for design and algorithms for actual processing. Our research has been mainly concerned with the latter classification and has addressed structural constraints which arise because of resource limits and/or a desire for efficiency. This has led in a number of problems to some algebraic decomposition of algorithms with the intent of achieving parallelism and quality, as measured by the degree of approximation (bias) and roundoff noise and similar errors (variance). More recently, we have been studying problems which involve both classes of algorithms; for example, what is a good design (filter function or window) which must conform to structural constraints dictated by processing considerations. This has proven to be a rich area of application.