Generalised dimension characterisation of radiotransmitter turn-on transients

Three different dimension measures have been tested for the characterisation of radio transmitter turn-on transient signals for a transmitter identification system. Fractal, information and correlation dimension trajectories were obtained through the application of a sliding window to the transient signal and resulting dimension values were used to construct feature vectors. Features were then classified by using a probabilistic neural network and experimental results of three classification systems were presented. The features estimated from information and correlation dimensions produced better classification results than a system using fractal dimensions.