Removal of Interference from FSK Signal Based on Independent Component Analysis

The phase of the frequency-shift keying (FSK) signal transmitted in frequency shift track circuits is continuous. The detection precision for FSK signals is restricted greatly due to the influence of interference signals. At present FSK signals are normally detected by using methods of wave filtering, spectrum analysis or signal transformation etc. However, it is more difficult to realize the demodulation of the FSK signal when its spectrum is similar to the interference signal´s spectrum. In this paper, by using method of independent component analysis (ICA), the blind signal separation of two FSK observation signals containing interference signal is implemented. In which the interference signal is AM signal with similar spectrum or random-pulse signal. Through computer simulations, it is confirmed that the interference signal can be separated effectively using this method. The upper and down side frequencies of FSK signals also can be detected. Therefore, a new removal method of interference from FSK signals is implemented.