Frequency Estimation Based on Gaussian Fitting in Frequency Domain

In this paper, the most typical resonator SAW (Surface Acoustic Wave) temperature sensor system is taken as an example. According to the characteristics of its echo signal, analyzed and compared the signal processing methods, the methods to improve the frequency resolution are proposed. In time-domain FFT (Fast Fourier Transform) method, due to the limited length of received signal, the frequency discrete sampling restriction exists. Then it will reduce the estimation accuracy of signal frequency. Rife algorithm improves the accuracy by interpolation, Grandke algorithm improves the accuracy by windowing. Both algorithms increase the amount of computation. At the same time the signal obtained may be very biased. In this paper, a frequency estimation algorithm based on Gaussian fitting in frequency domain is proposed. The corresponding echo signal is obtained by increasing the step frequency step by step on the excitation signal. The amplitude of the echo signal is accumulated in a certain period of time. Excitation signal with different frequencies and cumulative amplitude of echo signalare fitted by Gaussian fitting. From the Gaussian fitting, we can get the frequency. The experimental results show that this method does not need FFT transformation, the amount of computation is relatively small, and the accuracy is much higher than Rife algorithm and Grandke algorithm.