A Probabilistic Approach to Abstract Communication Models

The "Abstract Communication Model for Distributed Systems" outlines a general method for describing executable specifications for all message-based distributed communication systems. This model is based on the ASM (Abstract State Machine) paradigm. The model is highly non-deterministic and is intended to be so in order to preserve the generality of its applicability. The ACM effectively separates the communication logic from the application logic during high-level modeling and simulation of distributed systems. This model could be specialized and used for modeling and simulation during very early stages of system analysis and design. Unlike other network simulators such as ns-2, ACM is a light-weight plug-in; high-level model for the communication component in a distributed application. ACM enables analysis at a semantic level, which is desirable during early stages of system design and analysis. Heavy-weight network simulators are unsuitable for such purposes; they could be used in later stages for accurate and elaborate simulations of the system implemented. Hence, the two approaches target complementary aspects of computer network modeling. The model as such is not implementable. In this project, we propose a natural extension of the ASM paradigm, which involves probability. We specialize the ACM partially by introducing probabilistic transitions where possible. Consequently, the model becomes amenable to statistical analysis, simulation and reasoning.