Modeling and Analysis of Self-Reflection Based on Continuous State-Space Approximation of PEPA

With the application and popularization of autonomic computing in the field of aerospace exploration, large-scale database management and critical network control, existing self-reflection models based on natural language or diagram can not meet the requirements of analysis and verification. In this article, two kinds of self-reflection models, one is the static and the other is dynamic, capable of model checking and quantification analysis are separately built using performance evaluation process algebra (PEPA). Besides, huge state-spaces of models cause that the traditional Markov chains implied are hard to solve, thus we use an approach of continuous state-space approximation, a recent breakthrough in the analysis of stochastic process algebra, to generate ordinary differential system (ODE) from the PEPA model avoiding state-space explosion. By analyzing the ODEs, it is found that reducing the latency time of monitoring as well as shortening the length of execution instructions plays an important role in improving the performance of self-reflection.