Systematic Design and Programming of Signal Processors, Using Project Management Techniques

Jan Zeman, George S. Moschytz

Research output: Contribution to journalArticlepeer-review

13 Scopus citations


This paper presents a systematic method of designing and microprogramming fast signal processors (SP's) while optimally utilizing the inherent parallelism of a given algorithm. The method employs graph-theoretical concepts and scheduling techniques from “project management” and permits an easy evaluation of the following key design parameters: i) the lower bound on the algorithm duration (Tmp) for given speeds of the arithmetic components in a signal processor, ii) the minimum arithmetic hardware requirements necessary for the completion of the algorithm within the time Tmp, and iii) the optimum algorithm schedule and the corresponding minimum attainable duration Tat, when constraints on the available SP resources are imposed. In combination with the data-flow analysis and resource allocation, the scheduling procedure permits the influence of the SP architecture on the schedule to be modeled and the basic architectural features of the SP to be determined. The results also provide information for automatic microprogram generation and for the assessment and comparison of signal processor performance and algorithm speeds. The synthesis of architectures is based on an “ideal, data-flow driven signal processor” that is introduced in the paper. The proposed design approach is demonstrated on various digital-filter algorithms.

Original languageEnglish
Pages (from-to)1536-1549
Number of pages14
JournalIEEE Transactions on Acoustics, Speech, and Signal Processing
Issue number6
StatePublished - Dec 1983
Externally publishedYes


Dive into the research topics of 'Systematic Design and Programming of Signal Processors, Using Project Management Techniques'. Together they form a unique fingerprint.

Cite this