A real-time algorithm for signal detection based on autocorrelation at low SNR

In an electronic detecting system, how to capture non-cooperative signal at low SNR is always a challenging work. This paper puts forward a real-time algorithm named Shifted-N points-Correlation Algorithm for detecting LFM (Linear Frequency Modulation) pulse signal based on digital autocorrelation. Applying a gliding window of proper length to truncate infinite signal into finite length, then calculating the autocorrelation function of the signal in the window, choosing the N^th point as the current output, finally, an envelope which represents the position and intense of the signal can be obtained, while the noise can be eliminated according to its non-relativity. This is especially suitable for detecting weak signal with unknown parameters under the background of band-limited noise. Moreover, when the parameters of the receiving system have been determined, an optimum algorithm can be found for specified system based on the MSNR Criterion. Besides, it can achieve recursive operation, which is easy for hardware to apply.