A REAL TIME GENERAL PURPOSE SIGNAL PROCESSOR ARCHITECTURE.

Digital signal processing has many applications in the areas of signal, radar, speech and image processing and real time implementation requires a very high throughput rate. Various processor architectures have been investigated 1-6 which reveals that a special purpose processor appropriate to the algorithms should be designed in order to achieve high throughput rates. Recently research efforts 7-14 are directed towards the exploitation of parallelism in the algorithms and parallel computation of these algorithms. The objective of this work is to propose new concepts for high speed computation of real time general purpose signal processing algorithms. A novel architecture of a real time general purpose data flow signal processor (DFSP), based on the binary tree structure, is proposed for real time signal processing applications. The data flow signal processor exploits distributed, parallel, and pipeline processing approaches to achieve high throughput rates. The processor utilizes the residue number system (RNS) for high speed signal processing applications. The arithmetic operations in RNS can be performed via Random Access Memory (RAM) look up tables, and the execution time of any particular arithmetic operation is reduced to the access time of a RAM. The data flow signal processor is demonstrated to be suitable for performing recursive, non-recursive digital filtering and convolution operations. Finally the thesis describes various alternatives for programming the DFSP, and an interactive program environment is used to write application programs without knowing the internal architecture of the DFSP.Dept. of Electrical and Computer Engineering. Paper copy at Leddy Library: Theses & Major Papers - Basement, West Bldg. / Call Number: Thesis1985 .J353. Source: Dissertation Abstracts International, Volume: 46-02, Section: B, page: 0601. Thesis (Ph.D.)--University of Windsor (Canada), 1985