Internet Traffic Categories Demand Prediction to Support Dynamic QoS

Abstract

Categorized Internet protocol, and demand prediction based on usage behavior can offer substantial benefits to dynamic Quality of Service (QoS) for internet service providers (ISPs). The critical requirement for dynamic QoS is to classify and predict network traffic in the next control time interval. Traffic prediction, with the aid of traffic categories can utilize the network resources more efficiently and support Dynamic QoS to function appropriately. This paper proposes an internet traffic categories demand prediction framework using machine learning techniques to support dynamic QoS. In this framework, three algorithms are implemented at transformation stage to feed into machine learning algorithms and develops efficient prediction model to predict internet network traffic demand. Experimental results show that prediction accuracy of the model is 98.97% and is efficient and suitable to support real-world network traffic prediction.