Expert System for Automatic Analysis of Results of Network Simulation

Abstract

The simulation of communication networks is an important task in the process of network planning and optimization processes. Such methodologies assure a higher probability for networks to operate successfully under different critical conditions, which are difficult or unpractical to be tested in the real networks. Typical applications of such simulations are the simulation of military tactical radio networks. The manual analyses of network simulation results, is a very time-consuming task and requires expert knowledge to correctly interpret such results. This is a good motivation to develop the system for automatic analysis with expert knowledge, which will ease the process of mission planning and training. For such needs, we have developed the simulation methodology and tools, supported by the expert system, which are going to be presented in detail within this chapter. During this chapter, we will briefly introduce expert systems (further ES), Command and Control Information Systems used in NATO and known solutions to simulate such systems. The ES is defined as an intelligent computer program with a certain level of expert knowledge, which using procedures to solve exactly specified problems. All definitions for expert systems, in many books, are quite similar, and they describe the way such system includes a rigid range of expert (specialized) knowledge or research domain. Within this area, it is capable of creating intelligent decisions. This is some kind of imitation, where a system tries to capture behavior of skills. Using the acquired knowledge; a system can analyze input/output information, solve problems, and utilize utensil decisions within the problem domain. From this point of view, these systems cannot solve all kinds of problems, but they can solve well-known and deduced problems. It is stated in one of the references that expert systems are based on knowledge (Hart, 1998, pg. 7), respectively on an information handler base. Classification by Sauter places an expert system on the right side of the straight line, where we can find systems, which handle information. Expert systems are closely related to artificial intelligence methods. As a rule, they share quality and quantity information, probability theory, fuzzy set theory, and a number of arithmetic and logic rules, based on heuristic expectations.