Classification of low SNR transient signals using training with noise technique

In this study, the performance of a system that can identify radio transmitters from their turn-on transients was evaluated in noisy channels. It was observed that the classification performance of the system degrades in low SNR values and to improve the performance, a technique based on training with noise was proposed. The proposed technique was tested with experimental data and it was shown that the technique improves the performance significantly.