Technical Research Of Typical Signals Pattern Recognition Under Wireless Channel

Facing the current increasingly sophisticated communication systems, the technical research of signal modulation recognition is always a hot spot. Due to multipath propagation in wireless communication and signal power attenuation brought by the atmospheric, the traditional modulation recognition algorithms cannot adapt to poor channel environment, and also the recognition rate is very low. In order to overcome the effects of multipath propagation in wireless channel so that we could recognize the wireless signals more accurate, this paper presents a pattern recognition algorithm aimed at wireless channel which is using power spectrum estimation, high-order cumulants and signal instantaneous amplitude characteristics as signal characteristics based on the research of the modulation signal feature extraction and classification, and also presents two kinds of classification methods respectively are based on decision tree and based on SVM according to the acquired characteristic parameters. In addition, a GUI interface for algorithm is designed, and the algorithm has been verified and tested by signals collected in real environment, and can be applied to engineering practice well.The main work of this paper can be summarized as follows:1. A detailed analysis of typical communication signals’modulation principle and signal fading in wireless channel is given. The modulation principle of communication signals includes analog modulation and digital modulation. The propagation of signals in wireless channel includes fading and attenuation. In addition, the time domain waveforms of seven modulation signals including2ASK, BPSK, QPSK,16QAM, 2FSK, MSK and AM in the5dB SNR environment of wireless multipath channel transmission case and non-multipath channel transmission case are studied in this paper.2. The key technologies of wireless signal pattern recognition like characteristic parameters extraction and classification are studied. A method of signal characteristic parameters extraction based on power spectral estimation, high-order cumulants and signal instantaneous amplitude characteristics is proposed, according to the signal characteristic parameter extraction after synthesis, two methods of modulated signal classification and recognition based on decision tree and support vector machine is proposed using the two classifiers. Each method has its own merits, we should make a reasonable choice with the needs of the actual situation.3. On the basis of wireless signal pattern recognition algorithm, a user operation interface based on MATLAB GUI is designed, and a real environment for signal wireless transmission is built for testing and verifying this paper’s algorithm. With the summarization of recognition results and the comparison between algorithm under the ideal simulation channel and this paper’s algorithm, the result shows that the wireless signal pattern recognition algorithm proposed in this paper can remain high recognition rate in bad channel conditions, so it proved that this paper’s pattern recognition algorithm is correct and effective in practical engineering.