Parameters Estimation Methods Of Linear Frequency Signals And Sinusoidal Frequency Signals

Nonstational signals exist extensively in natural and human life. As nonstational signals,Linear Frequency Modulation (LFM) signals and Sinusoidal Frequency Modulation (SFM)signals are widely used in communication, radar, sonar, earthquake detection and biologicalmedication and other fields. Since typical Fourier Transform cannot reflect the time-varycharacteristic of nonstational signals, the time-frequency analysis methods become effectivetools to solve this problem and they have been successfully applied in many fields at present.First, this paper study on several parameter estimation methods of LFM signals, such asFractional Fourier Transform (FRFT), Discrete Chirp Fourier Transform (DCFT), MatchedFourier Transform (MFT) and Cubic Phase Transform (CPT), compare their performance witheach other by simulation.Then, utilize the well time-frequency energy concentration property and anti cross termcapability of LFM signal by first order Local Polynomial Fourier Transform (LPFT1), proposethe LFM signal parameters estimation method based on Two Step LPFT1-Fast FourierTransform (TSLPFT1-FFT). Simulation result shows that compare to common methods, theproposed method improve the parameters estimation accuracy at the same signal to noise ratio(SNR), and it also adapts to the multi-LFM signals cases.Last, since the time-frequency function of SFM signals are sinusoidal curves, and theLFM signal analysis methods donot fit them. To solve this problem, the Sinusoidal FrequencyModulation Transform (SFMT) is proposed. Theoretical analysis clues that the SFM signalhas apparently concentration characteristic by discrete SFMT, moreover simulation resultshows that the performance of the SFM signal parameters estimation method based onDSFMT closes to the Cramer-Rao lower bound.