Role of Machine Learning Applications in Communication Network's Traffic Classification to Manage TCP Congestion
The examination of packet flow in network application using port number and payload-based identification is important area of network surveillance and management. Machine learning is alternative approach to identify network applications using payload independent features inter-arrival time and packet length. The decision trees like Bayesian network, Naive Bayes, C4.5 Naive Bayes tree algorithm used to demonstrate the impact on performance of reduction of feature set. The computational performance based on algorithms for traffic classification differs from classifying accuracy alone.