Monotone response-time derivation for distributed execution ofrule-based programs

摘要

A key index of the performance of a rule-based program used innreal-time monitoring and control is its response time. We first extendnthe definition of response time of an EQL rule-based program forndistributed computation. To reduce the response time through distributedncomputation, we decompose an EQL program into disjoint modules. We thenndescribe a tool which computes the response-times of finite-state EQLnrule-based programs according to the imprecise computation paradigm,ni.e., this tool always yields a range which is monotonically tightenednas more time is spent in the computation. During the computation, a userncan interrupt the analyzer and get both an intermediate result which isna guaranteed bound and a bound-quality factor which quantifies thentightness of this result. Our approach uses fast textual analysis to getninitial bounds. It then performs a heuristic search by pruning thenstate-transition graph to improve the bound quality. Anprogram-decomposition technique for reducing the search effort is alsondiscussed. An analysis example on an EQL program involvingn2⁵⁹ states is presented