| Wideband radar can provide better range resolution and more information of the target,thus is an important direction for the development of modern radar. The research on targettracking with wideband waveform can remove the narrowband tracking waveform of theconventional radars and reduce the waste of radar energy, which is important to radarsystem design. In this thesis, the theory and algorithms are researched to take advantage ofwideband waveform and improve the tracking performance.Wideband waveform can provide more information of the target, which is helpful totarget tracking. However, high range resolution also results in the reduction insignal-to-noise ratio and more false alarms. In this thesis, an integrated detection-trackingalgorithm and an attribute aided tracking algorithm are developed to improve the trackingperformance. The integrated detection-tracking system improves the detection performanceby using the Bayesian detection which increases the connection between detector andtracker. Since the Bayesian detection cost is hard to determine, some parameters of thedetector are also hard to obtain. The attribute aided tracking algorithm improves thetracking performance by using the attribute measurements provided by the widebandwaveform. The assignment weights of the measurements are modified by the attributeprobability, so the association and tracking performance can be improved. The difficulty inthe usage of attributes is due to the uncertainty in attribute states and the unstability inattribute measurements.Our research in this paper is summarized as follows:1. To predict the performance of the tracker, an algorithm is developed to predict thetracking performance of PDAF with Bayesian detection (PDAF-BD). In the proposedalgorithm, the influence of false alarms and missed detections are characterized by theinformation reduction factor. On this basis, the track loss probability and tracking errorare predicted by iteratively computing the probability distribution of measurements andthe filtering covariance. The simulations show that the predicted performance isconsistent with the simulation result. The algorithm can be used to optimize the parameter of PDAF-BD.2. To improve the tracking performance, Bayesian detection is applied to the trackingsystem and an integrated detection-tracking algorithm is developed. The informationfeedback from the tracking module is used to reduce false alarms. To determine theparameter of the detector, the tracking performance for different detection parameters ispredicted online, then the detection parameter is dynamically optimized to minimizethe filtering error. According to the simulation result, the proposed algorithm canimprove the tracking performance of wideband radar.3. To take advantage of the attribute information provided by the wideband waveform, anEM-based attribute aided tracking algorithm is developed. The attribute characteristicsare described by the hidden Markov model (HMM), and the joint probabilistic modelof kinematic and attribute properties is derived. On this basis, an EM-based iterativealgorithm is developed. The kinematic and attribute information in multiple frames areused for association and filtering. Simulation results show that the proposed algorithmhas better performance when the attributes of targets are available. |
