Radio Transmitter Transient Identification Using Fractal and Neural Networks

In this paper, fast and reliable identification of radio transmitters based on an analysis of their turn-on transients is presented. Since the transients are unique, they are called the fingerprints of the corresponding radio transmitters. Major tasks in a radio transmitter identification system are: (i) separating a transient from the channel noise, (ii) extracting important features contained in the transient, and (iii) classifying the transient based on these features. We have developed a system capable of segmenting noise efficiently, using the variance fractal dimension trajectory [Kins94] [Shaw97], with a modified triggering technique which improves the segmentation consistency. A variance fractal amplification technique [Kins94] is used for feature extraction and enhancement. Multifractal modelling of transients based on their strange attractors is also discussed in this paper to investigate the suitability of such compact representations in transient classification [SuKS99a] [SuKS99b]. This is motivated by the observation that a transient signal contaminated by noise can be considered as an output from a chaotic dynamical system. The probabilistic neural network (PNN) is used in the classification stage. Preprocessing of the PNN using the principal component analysis (PCA) and the self-organizing feature map (SOFM) is performed to reduce the dimensionality of the PNN inputs and to cluster inputs, thus improving the training and classification speed. Experimental results have demonstrated that the radio transmitter identification system is faster than the previous implementations. More specifically, the training time for a network consisting of 400 transients can be reduced to about a half of the original training time, 241 seconds. The system can classify radio transmitters not only according to their manufacturers and models, but also serial numbers, with the average classification rate around 97%.