| Summary: | 碩士 === 中華科技大學 === 電子工程研究所碩士班 === 102 === Radar signal and speech recognition, the technology is quite mature, and the demand for more and more important. The principal purpose of this study was to use the linear predictive cepstral feature parameters which used in speech recognition and speaker identification, and use the Gaussian mixture model to FM transmitter signal recognition which model originally used in speaker recognition, face detection, and radar signal recognition. Using the received the FM transmitter signal, the parameters of the energy weighted center frequency, the characteristics obtained from the moments of power spectrum of the received signal, the moments of the received time domain signal, the linear prediction coefficients(LPC), and the linear prediction cepstrum coefficients(LPCC) were calculated, and then, the different combination sets of the calculated parameters were used to create different Gaussian mixture models(GMM) for the proposed FM transmitter signal recognition algorithms.
The different carrier frequencies, different modulation indexes and different SNR of FM transmitter signal were generated to be identified. The simulation results show that using the combination of the energy weighted center frequency, the characteristics obtained from the moments of power spectrum related characteristics and the linear predictive cepstral coefficients is better, and the recognition results show that the proposed method in this thesis works well.
|
|---|
