GMSK Modulated DSSS Signal Separation Based on Principal Component Analysis

Blind source separation (BSS) has been introduced in direct sequence spread spectrum (DSSS) system to suppress multiple access interference (MAI) without prior information. BSS or independent component analysis (ICA) based anti-interference receiver can separate the legal information sequence from the interference to improve the communication performance. However, the previous BSS or ICA based receiver is proposed for antipode CDMA signal where the information sequence is binary. Because of the continuous phase characteristics of GMSK, it is difficult for the BSS or ICA based receiver to achieve good performance in GMSK-DSSS systems. According to this, a principal component analysis (PCA) based receiver is proposed in this paper. Firstly, the waveform data of the signal is transformed into multi-channel signals. Secondly, for each channel signal, the covariance matrix is estimated. Then the covariance matrix is decomposed to obtain the eigenvalues and the eigenvectors. The un-mixing matrix is calculated with eigenvalues and eigenvectors. Finally, the information sequence of legal user is separated from the interference user with the un-mixing matrix. To verify the performance, simulation experiment is carried out. The Results indicate the proposed method outperforms the ICA or BSS based method.