Machine Learning Techniques for NTMA and VQA

This chapter delves into the machine learning (ML) techniques that surround and enable network traffic monitoring and analysis (NTMA) and video quality assessment (VQA). Through passive listening of network parameters’ being reported by the network interface, the Node.js server formulates a series of arrays that keep track of network traffic collected over time. The same applies to ML-derived mean opinion score (MOS) values through video streaming. Techniques such as multilayer perceptron (MLP) are used for regression—i.e., the prediction of both network traffic and video QoS scores—while the classification used is the K-nearest neighbors (KNN) model. This chapter contains a thorough mathematical breakdown of all the algorithms coded in this book for both classification and prediction of network traffic metrics, together with the production of an MOS score for video quality.