Improvement of transmitter identification system for low SNR transients

Performance of radio transmitter identification systems for low SNR transients can be improved by utilising the transient SNR information in the classifier training stage. The SNR of the unknown transient, which determines the amount of noise to be added into the neural network training set, has been estimated at the classification stage using pre- and post-trigger samples of the captured noisy transient signal. Performance improvement observed in simulations is verified with experimental data.