Generalization and Optimization of Feature Set for Accurate Identification of P2P Traffic in the Internet using Neural Network

摘要

P2P applications supposedly constitute a substantial proportion of today's Internet traffic. The ability to accurately identify different P2P applications in internet traffic is important to a broad range of network operations including application-specific traffic engineering, capacity planning, resource provisioning, service differentiation, etc. In this paper, we present a Neural Network approach that precisely identifies the P2P traffic using Multi-Layer Perceptron (MLP) neural network. It is general practice to reduce the cost of classification by reducing the number of features, utilizing some feature selection algorithm. The reduced feature set produced by such algorithms are highly data-dependent and are different for different data sets. Further the feature set produced from one data set does not yield good results when tried upon other data sets. We propose an optimum and universal set of features which is independent of training and test data sets. The proposed feature set has enabled us to achieve significant improvement in performance of the MLP classifier. The few features in the proposed feature set results in a significant reduction in training time, while maintaining the performance, thereby making this approach suitable for real-time implementation.