Compressive Frequency Hopping Signal Detection Using Spectral Kurtosis and Residual Signals

Over the last decade, the frequency hopping (FH) detection has attracted a great deal of interest due to its effectiveness. In general, FH adopts the concept of changing a carrier frequency following a random sequence which is only known by a transmitter and a receiver. In order to detect this unknown signal and determine its component of frequency hop, this project aims to take advantage of Compressive Sensing for sparse signal in frequency domain. Spectral kurtosis is firstly calculated to identify peak locations which represent hopping frequency. Then the most optimal peaks are selected by applying Compressive Sensing Match Pursuit algorithm to overcome noise uncertainty. The hop time is finally obtained from the residual between the given signal and the estimated one. The experiment results show that the system performs effectively even when SNR values are below -10 dB, and the presented approach can effectively detect the hopping time. Compared to other conventional approaches, our proposed method is appropriate for lower and more widely varying SNR and also considerably reduces the complexity.