Blind estimation algorithm for frequency hopping parameters based on enhanced time-frequency ridge extraction

In existing blind estimation algorithms for frequency-hopping parameters based on time-frequency ridges,performance degrades significantly when the signal-to-noise ratio(SNR) falls below-5 dB. To address this issue, an improved blind estimation method based on time-frequency ridges is proposed. First, the time-frequency matrix of the received signal is obtained using the short-time Fourier transform(STFT). Then, Gaussian filtering is applied for signal smoothing and noise suppression. Since the energy distribution of fixed-frequency interference differs from that of frequency-hopping signals in the time-frequency matrix, an energy cancellation method is used to eliminate such interference. Next, OTSU thresholding is applied to the matrix after energy cancellation to further remove residual fixed-frequency interference and noise, yielding a clearer time-frequency representation. Time-frequency ridges are then extracted from this image, and the least-squares method is employed to fit the hopping instants along these ridges, enabling accurate estimation of frequency-hopping parameters. Simulation results show that the proposed method effectively produces clean time-frequency images even at SNR levels as low as-5 dB in the presence of fixed-frequency interference. The average relative estimation error of each parameter remains below 1%, and the method demonstrates strong robustness against varying intensity and quantity of interference.