Clutter detection in pulse-doppler radar systems

Among various types of radar systems, the pulse-Doppler radar is the most widely used one, especially in military applications. Pulse Doppler radars have a primary objective to detect and estimate the range and the radial velocity of the targets. In order to have a basis for the detection, first reflected echo signals are matched filtered and then the time-alligned pulse returns are transformed to the Fourier domain to obtain the range-Doppler matrix. The resulting range-Doppler matrix is input to target detection algorithms. For this purpose, constant false alarm rate (CFAR) algorithms are run on the range-Doppler matrix. It is useful to run different CFAR algorithms inside the clutter region and outside the clutter region because the statistics are different inside and outside of the clutter. In order to achieve this discrimination, the position of the clutter has to be detected in the range-Doppler matrix. Moreover, the clutter may not always appear around zero Doppler frequency when realistic terrain models and moving platforms are considered. Two algorithms for clutter detection using range-Doppler matrix elements are investigated and their performance analysis is presented in this thesis. The first algorithm has higher error rates but lower computational complexity,whereas, the second one has lower error rates but higher computational complexity. Both algorithms detect clutter position by filtering the range-Doppler matrix elements via non-linear filters. In addition to the probabilistic error rate analysis, simulation results on some realistic cases are presented. It is concluded that the first algorithm is a good choice for low clutter-to-noise ratio values when a low-complexity algorithm is required. On the other hand, the second algorithm has better performance in all clutter-to-noise ratio values but it requires more computational power.