Spectral analysis of nonuniformly sampled data using a least square method for application in multiple PRF system

This paper addresses the problem of spectrum analysis from nonuniformly sampled data using the least squares method for applications in multiple PRF radar systems. The benefits of using a direct spectrum analysis using nonuniformly spaced data instead of using a FFT based CR (Chinese remainder) theorem is presented. Using a spectrum analysis based on nonuniformly spaced data, the blind bursts that the target frequency is folded into the clutter band can be overcome easily. Additional benefits of using an unevenly spaced data spectrum is its robustness to noise. It is shown that the least squares method outperforms conventional Chinese remainder theorem based on FFT method by 2.5 dB of SNR. This can also be an efficient method when there exist multiple interferences with same magnitude as the target. To obtain a spectrum from nonuniformly sampled data, the well known Lomb periodogram is modified to fit a complex sequence. The phase component is also recovered from the power representation to determine whether the target is opening or closing. The result of a modified scheme is compared to the result of the Lomb periodogram.