High-frequency Continue Wave (CW) telegraph has been an important tactical communication mode. However, this telegraph signal is often corrupted by impulsive disturbance and exhibits serious frequency deviation due to the poor High-frequency channel. As a result, the conventional fixed frequency tracking method based on Gaussian noise assumption may lose the tracking of time-varying CW signal. To enhance its automatic recognition and provide a reliable substitute for human effort, an automatic detection and recognition system is proposed. The system presented here first utilizes a robust extension of Kalman filter to approximate the time-varying CW telegraph waveform in the presence of impulsive interference. Then the Support Vector Machine (SVM) algorithm is applied to the recognition of CW telegraph with unstable code speed. Experimental results indicate that the proposed system could effectively suppress the impulsive noise and provide a high recognition rate for the time-varying CW signal.
[1]
Cheng-Hong Yang,et al.
Morse code recognition using support vector machines
,
2003,
IEEE EMBS Asian-Pacific Conference on Biomedical Engineering, 2003..
[2]
Jiang Yong,et al.
Automatic reception of High-frequency CW telegraph with Support Vector Machine
,
2010,
2010 2nd International Conference on Future Computer and Communication.
[3]
Sunghan Kim,et al.
Detecting and tracking tremor in spike trains using the rectangular model based extended Kalman smoother
,
2010,
Journal of Neuroscience Methods.
[4]
Zeng Li.
Adaptive Filtering of Weak High-Frequency CW Telegraph Signal
,
2010
.
[5]
Farhat Fnaiech,et al.
An Efficient DSP-FPGA-Based Real-Time Implementation Method of SVM Algorithms for an Indirect Matrix Converter
,
2011,
IEEE Transactions on Industrial Electronics.