Mode-Based Refactoring Method for Real-Time Software

"This project investigates migration of real-time software written in Ada to new platforms whose architectures are new and novel (pipeline machines, vector processors, etc.) and who support other languages (C++, real-time Java, etc.). The approach beinginvestigated refactors legacy programs to optimize their real-time performance under stressed operating conditions using mode segmentation concepts instead of object or functional recovery methods. The technique being explored consists of three parts.First, the recursive functions in the program are combined to create one big function. Then, this function is partitioned into Unix-like processes using cluster analysis such that each grouping performs a specific mode with predictable behavior. Thirdand finally, the segment's behavior is examined and optimized using factor analysis to address any architectural, operational and environmental constraints using a library of proven algorithms. Our approach offers significant advantages over the moreclassical approaches. Since the program is segmented into modes, it can be refactored easily to run on new architectures at minimal cost without any change in behavior and significant reduction in run-time performance. Firms concerned about real-timeperformance of legacy code can exploit this technique to migrate existing real-time applications at minimum cost to new platforms and languages. The need for such migration support is broad and