A Fast, Partially Parallelizable Algorithm for Predicting Execution Time of EQL Rule-Based Programs

Abstract

Real-time expert systems are embedded decision systems which must respond to changes in the environments within stringent timing constraints. A major problem impeding the use of rule-based expert systems in real-time environments is the difficulty in predicting the response time of these rule-based systems. In this paper, we tackle this problem with a fast, partially parallelizable response time analysis algorithm for a class of EQL rule-based programs with constant assignments in the action parts of the rules.