FM Signal Feature Extraction Using FFT-Based Integrated Interpolation Frequency Estimator

Feature extraction of FM signal is the first and most important step in modern automatic communication signal recognition. Based on the real-world collected silent FM observation, a systematic method for feature extraction of FM signal is presented after an intensive experiment modeling analysis, which includes Hilbert Transform and the key robust single-tone parameter estimation technique---FFT-based Integrated Interpolation Frequency Estimator (IIFE) which integrates the contribution of more FFT coefficients than traditional FFT-based estimators in single tone frequency estimation through a novel normalized frequency offset estimation moving average model and Maximum Likelihood estimation. Computer Monte-Caro simulations show the performance of estimator approximates the CRLB closely even at as low SNR as -9dB at 256 FFT points with little addition of computation time compared to traditional FFT-based ones and that the proposed IIFE is effective in its application to feature extraction of FM signal.