Automatic reception of High-frequency CW telegraph with Support Vector Machine

High-frequency CW telegraph has been an important tactical communication mode. The reception of CW telegraph primarily requires manual work. Thus, its accuracy is easily affected by human physiology. To improve its reception efficiency and provide a partial substitute for human effort, an automatic detection and recognition system was proposed. The system presented here utilizes Kalman filtering algorithm to extract the time domain characteristics of CW signals and Support Vector Machine to deal with unstable code speed. Experimental results revealed that the proposed system generated high recognition rate.