Abstract

Influence control is a very challenging issue indynamic reconfiguration and still not well addressed in theliterature. This paper argues that dynamic reconfigurationinfluences system execution in four ways: functional update,functional side-effect, logical influence on performance andphysical influence on performance. Methods includingversion control, transaction tracing, switchingreconfiguration plan, and reconfiguration scheduling havebeen proposed for controlling the influence. These methodsare integrated into the Reconfigurable Data Flow (RDF)model, which is designed to support the dynamicreconfiguration of stateless data flow systems. The RDFplatform is an implementation of the RDF model on Javaplatform. The RDF platform is implemented as an openframe for different reconfiguration planning algorithms andscheduling policies to be simulated and their influence to bequantitatively compared on a single platform. Using theData Encryption and Digital Signature System as a casestudy, tests have been done on the RDF platform to examinethe influence of different reconfiguration planningalgorithms and scheduling policies. Experimental resultsshow that the methods proposed in this paper is effective incontrolling the influence of dynamic reconfigurations.