Performance Estimation of Encrypted Video Streaming in Light of End-User Playback-Related Interactions

Abstract

Our research will look into realistic end-user service usage behavior patterns and their corresponding implications on the in-network Quality of Experience (QoE) monitoring for HTTP adaptive video streaming (HAS) services in wireless and mobile networks. The main goal is to establish a methodology for developing and testing machine learning (ML) models for estimating end-user QoE-related Key Performance Indicators (KPIs) in the context of user-initiated playback interactions. The initial phase will be to investigate user behavior when utilizing video streaming services on mobile devices and propose a user interaction model. In addition, a methodology for automated data collecting, processing, and analysis will be created, which will include the creation of a framework that combines user interaction simulation based on the proposed model. Extensive experiments will be carried out to train ML models for KPI estimation, and the resultant KPI estimation models will be evaluated. This paper presents a current state-of-the-art review of the corresponding topics, as well as the current state of our research and preliminary findings.