Implement Of Track-before-detect Algorithms Based On Real Radar Echo Data

With the fast development and wide utilization of stealth technology, the signal-noise-ratio(SNR) of targets decrease, which imposes a great challenge to radar’s detection and tracking ability. In a conventional detect-before-track(DBT) scheme, target detection is performed in each frame using a threshold operation. Then target tracking is achieved through the following data association processing. Considering with weak targets, however, the performance of the conventional DBT is decreased shapely. Recently, much attention has been raised about the track-before-detect(TBD) technique, where detection is achieved by jointly processing multiple measurements. Since useful information is not discarded at earlier stages, the TBD technique has potential to achieve a high performance. At the same time, since multiple measurements need to be processed simultaneously, the signal processing algorithm is crucial to the overall performance.In this dissertation, TBD algorithm for weak targets in radar system is investigated. The main results are as follows:1. The standard Hough transform to solve the TBD problem is first introduced by assuming that the target is moving towards the radar. Next, the 3D Hough transform to a general target detection problem is discussed. Their basic principles, simulation analysis and processing are discussed in details. Simulation results are provided to demonstrate the performance.2. The particle filtering technology to form is introduced, and a basic TBD model is established. In this thesis, a standard and an improved version of particle filter based TBD algorithms are also introduced. Then, simulation results for a uniform linear motion scenario and a coordinated turn motion are provided. The results show that the improved particle filter-TBD algorithm outperforms the standard algorithm slightly in a low SNR.3. In order to further validate the usefulness of Hough transform and particle filter in TBD applications, those two TBD techniques are discussed with real radar data. Thedata analysis and processing are explained in detail, as well as the detection results. Ourresults show that both the Hough transform TBD and the particle filter TBD algorithmshave a better detection performance than the conventional method. The advantage of theHough transform based algorithms is its simplicity. However, the application of Houghtransform based algorithms is limited to scenarios where the target should move in astraight line. In contrast, the particle filter based TBD algorithm is applicable to moregeneral applications. However, the implementation is more complicated.In this thesis, Hough transform and particle filter TBD techniques about their basic principles, simulation analysis and processing are discussed in detail. Then those two TBD techniques are applied to the real radar data to validate the usefulness of them in low SNR.