Semantic-driven Performance Evaluation

We propose a structural operational semantics for mobile and distributed agents. From it we derive a stochastic transition system labelled by actions and their costs. These costs reflect the (net) architecture on which agents run. We then map stochastic transition systems to Markov chains, and performance evaluation is carried out using standard tools. The results given by our approach are shown to agree with those obtained via classical evaluation techniques on a case study involving mobile computation. 1 Introduction Recently, stochastic process algebras [11, 12, 3, 5, 16] have been proposed as a mean to specify software and to derive general performance measures. Process algebra specifications are built by combining the basic actions, often called prefixes, that a system performs. The occurrence of a prefix is represented by a transition labelled by the associated action. The stochastic variants of process algebras associate probabilistic distributions with prefixes. The transition..