Study On Target Tracking With Cognitive Radar

With the development of radar technology, the radar surveillance scene has become more and more complicated. And the target tracking in complex environment is a great challenge for modern radar. The traditional active radar(TAR) only adopts adaptive filtering in the receiver. Then, the fore-active radar(FAR) is presented to adopt the adaptive technology in the transmitter as well. As a result, the traditional open-loop signal processing system become a closed-loop system. As the next new generation of radar, cognitive radar(CR) aims to add knowledge assistance to the FAR and enables radar with more intelligence in order to enhance the target tracking performance.In this thesis, the basic problem of target tracking is studied under the frame of CR. The main work and contributions are summarized as follows:1. The concept of CR, including its basic structure, component and mathematical model is studied.2. The workflow of the perception-action cycle(PAC) is studied. The relationship between the transmitted wave parameters and the measurement noise covariance is studied. Then, the adaptive filtering under different transmitted waveform parameters is deduced. The methods of design waveform library as well as the dynamic programming is also studied. Compared to the traditional signal processing methods, PAC can improve tracking precision.3. The design idea of CR’s memory which is constructed using neural network is studied. Furthermore, the issue of how the memory cooperates with PAC is studied. The results show that CR’s tracking performance is better than FAR’s.4. In order to solve the tracking problem of CR in clutter environment, we present to introduce the nearest neighbor filter and PDA into the perceptual memory, respectively. Experimental results demonstrate the effectiveness of the presented methods.5.Through three kinds of different tracking environments, with and without clutter, the performance of CR is compared to those of FAR and TAR. Experimental results show the tracking performance of CR is better than FAR, and much better than TAR.