THE DESIGN AND IMPLEMENTATION OF A MULTIAGENT INFORMATION FUSION SYSTEM FOR SUPPORTING DECISION MAKING IN AN AIR OPERATION PLANNING

• Main Author: DATUMAYA WAHYUDI SUMARI (NIM 23206008), ARWIN
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/7327
• Institution: Institut Teknologi Bandung

In a dynamic Air Operation (AO), the quality of the information which is sent to the decision maker gives a significant impact to the combat strategic planning in order to win the war and reduce the lost. A fast and correct decision making is depended on the accuracy and the speed of processing of the information gained from the distributed and strategically located sensors. In the AO, the data gained from intelligent activities such as surveillance and reconnaissance which is kept in an intelligent data base, will be processed and analyzed from intelligent, operation, personnel, logistics, and communication electronics perspectives. The analyzed data then will be inferred and presented to the decision maker. <br /> <br /> <br /> <br /> Information that is flown continuously could cause an information explosion. Another problem is the process of data analyzing still done in conventional ways so that it is not realistic to cope with the dynamic of the AO in this information age. Those problems impact to the Observe, Orient, Decide, and Act (OODA) decision cycle speed reduction. In order to face the problems, a system that is capable to combine the data from the sensors and the information from the analyzed data to gain complete and accurate information in a speedily fashion is proposed, designed, and implemented in this thesis. Such a system is called as a MultiAgent Information Fusion System (MAIFS). <br /> <br /> <br /> <br /> The fusion information process uses the combination of the Dasarathy process model at Data In-Data Out (DAI-DAO), Feature In-Decision Out (FEI-DEO), and Decision In-Decision Out (DEI-DEO) levels, and the Joint Director’s Laboratory (JDL) process model at level 0, 1, 2, and 3. For MAIFS implementation purpose, the Dasarathy process model is generalized by modifying the FEI-DEO and DEI-DEO levels to FEI-Inference Out (IEO) and Inference In-Inference Out (IEI-IEO) levels. The generalization is also done to the JDL process model by defining level 0 for sub-feature assessment, level 1 for feature assessment, level 2 for situation assessment, and level 3 for whole situation assessment. <br /> <br /> <br /> <br /> The information fusion mechanism will be applied by using the Maximum Score of the Total Sum of Joint Probabilities (MSJP) as the primary method which is a generalization of the Bayesian method, the voting method, the Boolean OR method, and the classification method and will be done by the Information Fusion Agent (IFA) according to the processed input and the desired output forms. This information fusion mechanism forms a hierarchical MAIFS architecture that consists of level 0, 1, 2, and 3 based on the kind of information fusion produced. A main agent is defined to fuse the information fusion products of the local agents in order to gain complete information that covers the whole information fusion products. The MAIFS information fusion and information processing are implemented by using the C++ language programming for a case study validation on the data taken from an intelligent database of Indonesian Air Force Olah Yudha AO training which has been obscured. <br /> <br /> <br /> <br /> The result of validation shows that MAIFS is superior over the conventional system viewed from the aspect speed of data processing from the beginning to information presentation to the decision maker. This superiority directly affects to the OODA decision cycle reduction so that the decision can be decided quickly. The MAIFS primary superiority is the fusion information product that is displayed in graphics form about the situation of the operation areas. From the presented information, knowledge can be extracted and give a belief and certainty to the commander to quickly decide the most proper AO strategy to be done with a minimal lost. <br />