Document Type : Original Article


Egyptian Armed Forces.


Moving Target Detection (MTD) is an automated radar signal and data processing system, which is designed to improve the performance of radar systems in the presence of various forms of clutter. Consequently, it provides high probability of detection (Pd) for an acceptable probability of false alarm (Pia). It employs coherent, linear Doppler filtering, adaptive thresholding and a fine ground clutter map to reject ground clutter, rain clutter, birds, and interference. The current MTDs relay on the Fast Fourier Transform (FFT) of the total received data sequence to estimate the clutter power spectrum and consequently reduce or remove its effect on the detection performance. Direct FFT leads to high sidelobes level and requires a significant large computation time. The high level sidelobes increases the false alarm probability at the output of the Doppler filters bank. A solution for reducing the effect of spectral sidelobes is the utilization of window functions. However, this solution leads to widening the main spectral lobe and reduce the Doppler resolution. Also it increases the hardware complexity of the system. In the present work, Bartlett method for spectral estimation which depends on dividing the received data sequence into a number, K, of nonoverlapping segments and averaging the calculated FFT for each segment over K, is applied in the MTD instead of the direct FFT for typical ground based radar. The proposed method enhances the target detection capabilities, by providing higher detection probabilities, lower false alarm rates and an additional gain of 7-10 dB in the improvement factor, in the presence of ground and weather clutter, compared to the traditional one. This is because the sidelobe levels obtained are very small in magnitude. This in turns facilitates the realization of the Doppler filters bank without using additional weighting. The obtained performance can be achieved by applying the direct FFT with weighting function to the total received data sequence, which leads to a more hardware complexity and long time calculations compared to the proposed method. Computer simulation results are presented to support the superiority of the proposed technique.