Discrete event system specification (DEVS) distributed object computing (DOC) modeling and simulation

Abstract

This research examines an approach to modeling and simulating distributed object computing (DOC) systems as a set of discrete software components mapped onto a set of networked processing nodes. Our overall modeling approach has clearly separated hardware and software components enabling systems level, distributed co-design engineering. The distributed co-design engineering refers to a formal approach to concurrent hardware and software systems engineering that provides a tractable method for analyzing the inherent complexities that arise in distributed systems. The software abstraction forms a distributed cooperative object (DCO) model to represent interacting software objects. The hardware abstraction forms a loosely coupled network (LCN) model of processing nodes, network gates, and interconnecting communication links. The distribution of DCO software across LCN processors forms an object system mapping (OSM). The OSM provides a sufficient specification to allow simulation investigations. During simulation, the behavioral dynamics of the interacting DCO software components "load" the LCN processing and networking components in terms of memory utilization, computational demands, and communications traffic. The resource constraints of the LCN components, likewise, impose performance constraints on the associated software objects. Class models of the DCO, LCN, and OSM component structures and behavior dynamics were formally developed using the Discrete Event System Specification (DEVS) formalism. These class model specifications were implemented in DEVSJAVA, a Java implementation of DEVS. Class models of experimental frame components were also developed and implemented to facilitate analysis of individual DCO and LCN components, as well as interdependent system behaviors, during simulations. The resulting DEVS-DOC environment enables distributed systems architects, integration engineers, and automated system designers to conduct performance engineering and trade-off analysis of distributed system structures, topologies, and technologies. This research utilized the resulting DEVS-DOC environment in four case studies. These case studies demonstrate the structural independence and behavioral interdependence of the hardware and software abstractions, the ability to model and simulate real world systems, and the complex interactions that arise in distributed systems can be methodically analyzed.