Multifractal characterization for classification of telecommunications traffic

A novel multifractal approach to telecommunications traffic classification is presented as an improvement over traditional traffic classifiers. The fundamental advantages of using multifractal measures include normalization and a very high compression ratio of a signature of the traffic, thus leading to more reliable and faster implementations, and the ability to add new traffic classes without redesigning the traffic classifier. The variance fractal dimension trajectory is used to provide a multifractal "signature" for each type of traffic over its duration. As a multifractal, the Renyi dimension spectrum is constructed to show the unique characteristics of each type of traffic. A probabilistic neural network is trained with the variance fractal dimension trajectory of the traffic, and has demonstrated 90.7% classification accuracy with a 585:1 compression ratio, and 90.4% classification accuracy with a 1024:1 compression ratio.