Читать книгу The Smart Cyber Ecosystem for Sustainable Development - Группа авторов - Страница 59

QoS parameters are normally used by network administrators to assess the network performance. The parameters include throughput, loss rate, delay, and jitter. However, QoE is a parameter used to represent the user perception and satisfaction of the services. Developing prediction methods for QoS and QoE parameters helps network operators and service providers to offer high quality services [13]. SDN has been used to facilitate the implementation of different algorithms for QoS/QoE prediction [36–39].

The authors of [36] propose a linear regression ML algorithm for QoS prediction in SDN-based networks. A decision tree approach is used to detect relations between KPIs and QoS parameters. The authors show that the method can predict congestion and thus provide recommendations on QoS improvement. The researchers in [37] utilize two ML techniques for estimating QoS parameters for video on demand applications.

QoE prediction was addressed in [38–39]. The method of [38] was designed for video streaming in an SDN-based network, where QoS parameters are employed to estimate the mean opinion score. The SDN controller is used to adjust video parameters to improve QoE. In [39], the authors use neural network and KNN algorithms for predicting QoE parameters using video quality parameters.