Comparison of nearest neighbor and probabilistic data association methods for non-linear target tracking data association

Abstract

Abstract —Target tracking problems are theoretically interesting, because the origins of the measurements are not identified. Data association is one of the key techniques on tracking with radar. The problem of data association for target tracking in a cluttered environment with linear target model and non-linear measurement model will be discussed. Firstly, evidences are constructed based on spherical coordinates. Then, the association decisions are constructed according to nearest neighbor and probabilistic data association methods. The simulation results show that the latter method has better performance than the former. Moreover, the results will be compared to linear target tracking, which is really common in data association techniques and it will be shown that there will be a slight decrease in performance of target tracking with non-linear measurement model. Keywords— Target tracking; Data association; Nearest neighbor; Probabilistic data association I. I NTRODUCTION owadays with the development of modern system, radars has developed to detect and track air objects. The output measurements of radars are angular measurements of target track. In data association process, measurements are divided into two parts, known tracks and false alarms. Thus, accurate state estimates of tracks can be updated. A key problem in target tracking is the hit-to-track data association. The aim is to estimate some unknown parameters expressed in terms of the state vector and its associated covariance matrix. In order to solve this problem, there are two kinds of conventional methods to associate measurement-to-track, which are Nearest Neighbor Data Association (NNDA) and Probabilistic Data Association (PDA). The measurement that is closest to the predicted target-oriented measurement is known as the nearest neighbor (NN) measurement in target tracking. One of the most common and widely used methods for tracking in clutter is NN which only uses the NN measurement as if it were the one originated from the target of interest. This filter can be used either alone or as a module in a more complex algorithm. [1], [2] and [3]. The PDA is a method for associating hits where only one target is available. It assumes that all hits are in a particular target extension gate and originated either from a target or random clutter. If another target is persistently in this target extension gate, the results are poor and may be wrong.[1], [2] and [4]. References [3], [4], [5] and [6] discussed the data association problems of target tracking with assumption that target and measurement models are linear, and kalman filtering is used. [7] and [8] Here we discuss the problem of data association with non-linear measurement model for radars. Thus NN and PDA methods should be modified in order to be used for non-linear systems. As number of targets should be known before tracking, we consider the false alarm as clutter. Moreover, the constant velocity model is used in [3] and [4], but here we used a more complicated singer mode for target movement. Dynamic models of target movements are described in [9]. A validation gate is considered to omit the measurements with low probability of association. [10] Section II of this paper deals with problem description. Sections III and IV respectively deals with a short description of NN and PDAF algorithms including their physical constraints, requirements and formulation. Section V emphasizes some simulation results where the performances of elaborated NN and PDA are compared together and compared to linear NN and PDA. Finally, in section VI conclusion are discussed. II. P