Multisensor multitarget tracking in a cluttered environment

The performance of the tracking system is pivotal on the data association technique. The latter arises from the fact that: in practical systems where there is uncertainty of the measurement origin, such as tracking a target in presence of clutter, the effectiveness of the data association technique is vital. At low signal-to-noise ratio (SNR), it is very difficult to achieve proper data association. The Bayesian approaches, such as the probabilistic data association (PDA), does not yield acceptable performance because the target originated measurements may fall outside the validation gate. The objectives of the thesis are to analyze the conventional tracking techniques and to identify the areas where the fuzzy logic and the Viterbi-based techniques can enhance the performance of the overall tracking system in a cluttered multisensor environment.; Due to the non-target interference and/or high level of background noise in a highly cluttered environment, the task of distinguishing the target signal from any other signal is non-trivial, therefore, a fuzzy logic based technique is highly preferable. By applying the fuzzy logic technique to target tracking, we will be able to create a robust tracker that is capable of maintaining signal lock on a target at low SNR. The proposed fuzzy logic-based tracking algorithm is exercised against simulated as well as realistic data sets where the latter has been provided by Defense Research Development Canada-Atlantic (DRDC-Atlantic). Moreover, the Fuzzy tracker is combined with the interacting multiple model algorithm to track highly maneuvering targets in clutter. The results demonstrated that the fuzzy logic-based tracker yields superior performance than those of the conventional methods at low SNR situations.; The Viterbi data association (VDA) technique determines the most-likely state transition sequence in a state diagram. When the Euclidean distance is used as a distance measure, the VDA is the optimal maximum-likelihood detection method. In this thesis, an observation-based Viterbi data association (OVDA) technique has been further analyzed both analytically and through simulations. The algorithm shows promise in terms of its low computational load, thus it is a good candidate for the target tracking applications. In addition, the VDA algorithm is extended to tracking of multiple targets.; The thesis also presents an overview of the integration, registration, association, and fusion issues in a network centric environment. A data fusion architecture is proposed which combines most of the advantages of previously published architectures. The proposed architecture is analyzed from the points of view of data association and sensor fusion. Situation assessment, threat assessment, and sensor management are beyond the scope of this thesis and should be considered as future research activities.