An empirical investigation into CDMA network traffic classification based on feature selection

With the rapid development of CDMA systems, mobile network based applications have been undergone a tremendous growth in the past several years. The ability to accurately classify network traffic is of critical importance to the network design, troubleshooting, performance evaluation, and optimization. In this paper, we explore the design of an accurate and scalable machine learning (ML) based traffic classification system upon correlation-based feature selection (CFS) methods. With extensive data collected from a Tier 1 production cellular network, we experimentally show that our proposal achieves a high classification accuracy and low computational complexity.