Signal Processing Applications of Systolic Array Technology,
Abstract
Architectures and algorithms are examined for exploiting the large number of degrees of freedom available in current VLSI and projected VHSIC integrated circuit technologies to provide real-time implementations for sonar signal processing tasks. Such real-time signal processing tasks present a heavy computational load in the form of linear and multilinear transforms including beamforming and cross ambiguity calculation; matrix multiplication for covariance matrix estimation, least squares solutions for adaptive processing, and eigensystem solution for high resolution direction finding. The large number of degrees of freedom will frequently require not only parallelism in the computation, but also that special attention to be paid to the organization of the data flow, memory architecture, and propagation of required control information. This paper examines the matrix computation requirements for signal processing tasks including adaptive filtering, ambiguity function computation, recent methods of spectrum analysis and direction finding. Parallel processing architectures for implementating these computations are compared, with emphasis on recent developments in systolic architectures providing modular parallelism, local interconnects, and regular data flow for the major signal processing computations. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1983
- Accession Number
- ADP002601
Entities
People
- H. J. Whitehouse
- J. M. Speiser
- K. Bromley