PROBABILISTIC DATA ASSOCIATION BASED MULTIPLE TARGET TACKING

Abstract

Three important problems have been discussed in this thesis in detail. First, the problem of tracking is considered when poor and noisy measurements at receiver are available. Secondly, there is a need for predicting target location when observations are not available for long time i.e. when targets are not illuminated by the radar antenna. And thirdly, the most important problem of tracking single or multiple targets in the presence of clutter using PDAF or JPDAF algorithm respectively is discussed. Special problem of data outage for long time may arise due to limited resources, e.g. an Omni‐directional antenna rotating 3600 illuminates target for small interval of time, during illumination, any optimal or suboptimal estimator may be used to mitigate the noise and estimate a fine three dimensional position of the target. But when the target is out of sight as discussed in chapter 2 of this thesis, prediction technique must be applied for speculating target three dimensional location. Computer simulation results regarding above mentioned three problems including the special case of long time data outage are shown in this thesis. Three dimensional tracking of single and multiple targets in the clutter via PDAF and JPDAF respectively is one of the major works in this thesis. Moreover this thesis gives comprehensive details on detection theory, parameter estimation, state‐space recursive LS, Kalman Filter, estimation of states in the existence of data association uncertainty, pure MMSE approach, MMSE‐MAP approach, MMSE‐maximum likelihood approach, few other heuristic approaches, data association in non‐linear dynamic systems, some practical estimators, PDAF and JPDAF.