Expert system for analysis of tactical radio networks simulations

Now days the use of simulation tools rapidly increases because they are quite precise and gives satisfactory results, which are comparable with measured results on real communication networks. Most of available simulation tools have possibilities to present final results in graphical form. Such approach is suitable in cases, where expert user operate with small number of graphs and simulated scenarios. In cases where must user compare more than two individual graphs, analysis became complex and time voracious. We can decide that manual analysis of graphical results takes-up a lot of precious time, especially in cases of simultaneously analysis of multiple graphs. This time is economized by our system. Expert system (ES in the following explanation) is defined as intelligent computer program which includes certain level of expert knowledge. Such knowledge is stored in knowledge data base. Knowledge data base quality is one of the most important factors for such systems. It is some kind of function concerning data base dimensions and knowledge quality. A wide-spread base with high expert knowledge leads to a high-performance expert system. Knowledge must be stored into the base in the right format, because an expert system has to understand this knowledge, to create right decisions based on that knowledge. Detailed description of knowledge base is given in section 2.1. Some of the most important parts of ES are user interface, reasoning mechanism and fuzzy set for sorting and arranging simulation values into proper classes. Reasoning mechanism is one of the main parts of an expert system. It controls the operation of the whole ES. The mechanism must actively use the knowledge base for dealing with data, coming into the system, and for the derivation of suitable facts. The mechanism is composed of inquiring and reasoning processes, which helps in the solution search process. The most useful reasoning methods in the cases when we want to derive knowledge using production rules are: forward reasoning and backwards reasoning. Detailed description of reasoning mechanism is given in section 2.2. Fuzzy sets as main part of ES are generalization of regular crisp sets / 1, 2/. Meanwhile, the appurtenance function of a crisp set has a stock value (0, 1) (specific element belongs or does not belong to this set) the appurtenance function of a fuzzy set (mu(A)) has a stock value within the interval [0, 1]. We can reason, that a specific element in fuzzy set is contained by appurtenance, which is epsilon[0, 1]. For example we observe OPNET simulation graph for tactical radio received power data. For this data we define set A=(x; data in x is acceptable). Such a set contains all acceptable data. If we look at this set as an ordinary binary set, we can specify that data fully belongs to it or even does not fully belongs to it (two possibilities). A problem appears about 'acceptability' definition. In regular sets, passages between appurtenance and nonappurtenance are sharp (discrete). Passages between appurtenance and nonappurtenance in fuzzy sets are soft, slow and continuous (section 2.3). Fuzzy set is in our case used for analyzing results from transmitter utilization statistics and packet delay statistics, meanwhile radio visibility and message completion rate uses approach based on direct comparison between sent packets and received packets between communication units. Section 2.8 describes expert system user interface and manners to present results. Section four concludes the paper.