Akademik Veri Yönetim Sistemi
Tezin Türü: Yüksek Lisans
Tezin Yürütüldüğü Kurum: Orta Doğu Teknik Üniversitesi, Mühendislik Fakültesi, Elektrik ve Elektronik Mühendisliği Bölümü, Türkiye
Tezin Onay Tarihi: 2014
Tezin Dili: İngilizce
Öğrenci: Hasan İhsan Turhan
Danışman: MÜBECCEL DEMİREKLER
Açık Arşiv Koleksiyonu: AVESİS Açık Erişim Koleksiyonu
Özet:The aim of this thesis is to study attribute data fusion and decision making for targets tracked by a sensor network consisting of several radars. As an application deciding both target class and identity are studied. Since only partial information is available, Dempster-Shafer theory is used for this application to assign and combine probability masses. In this study, we focus on the problems of basic probability assignment and decision/data fusion. Classification of air vehicles according to their type is studied using the kinematic features obtained while tracking. The probability masses are obtained from tracker data and prior information that belong to possible target types. Prior information is modeled as a Gaussian mixture probability density function, while tracker data is modeled as a single Gaussian. This new methodology is tested with real data and its performance is examined by comparing it with the most similar method existing in the literature. Special to this type of air vehicle classification problem, a decision fusion approach is proposed that uses Bayesian formalism. The main difference of the proposed methodology from the existing methods is fusing the data before assigning the basic probabilities. Methodology is tested with real data and compared with the existing combination rules in the literature. Target identification is the decision of whether a target is a friend, hostile or neutral. This decision is made by using IFF Mod-4 information, IFF Mod-3 information, restricted area breach information, air corridor usage information and human-eye identification information. These piece of information are converted into probability masses and combined by using Analytic Hierarchy Process Interrogation methods and Dempster-Shafer Theory. Methodology is tested by using artificial scenarios.