Performance and reliability analysis of relevance filtering for scalable distributed interactive simulation

Abstract

Achieving the real-time linkage among multiple, geographically-distant, local area networks that support distributed interactive simulation (DIS) requires tremendous bandwidth and communication resources. Today, meeting the bandwidth and communication requirements of DIS is one of the major challenges facing the design and implementation of large scale DIS training exercises. In this article, we discuss the DIS scalability problem, briefly overview the major bandwidth reduction techniques currently being investigated and implemented in contemporary DIS systems, and present a detailed analysis on the performance and reliability of relevance filtering—a promising technique to improve the scalability of distributed simulation. The idea of relevance filtering is to analyze the semantic contents of the state update messages of a simulated entity (vehicle) and transmit only the ones found to be relevant to other entities. We present our entity-based model for relevance filtering and discuss the implementation of filtering-at-transmission and filtering-at-reception. We introduce the concept of filtering reliability and present different methods to eliminate or reduce filtering errors. Methods that can ensure complete filtering reliability while providing significant bandwidth reduction are developed. Performance evaluation results of relevance filtering and of the filtering reliability methods are presented. The insight gained from our work and the challenges still facing the design of large scale DIS training exercises are discussed.