Abstract:

As large instantaneous-bandwidth reconnaissance receiver may miss weak signals when intercepting signals corrupted by non-ideal noise, a novel method of signal detection based on morphological filter is proposed to process the received spectrum as one-dimensional gray-scale image. Noise floor is estimated with a morphological filter to correct the Welch spectrum. A detection threshold is obtained from the corrected spectrum using the constant false alarm rate processing, and signals are detected by comparing the corrected spectrum with the threshold. Simulation results show that the noise floor can be estimated and weak signals reliably detected in a given non-flat noise environment. The proposed method is suitable for hardware implementation due to the low computational complexity.

Key words: radio intelligence, signal detection, noise floor estimation, morphological filter