AN INTELLIGENT SYSTEM FOR PREDICTING VIDEO STREAMING QUALITY OF SERVICE

Abstract

A primary concern associated with live streaming systems has been the significant delay in waiting time. Typically, a new user will abandon a server before receiving service due to the waiting time, startup delay, or other types of delays exceeding their tolerance. To address these issues, the optimal approach is to possess the ability to precisely forecast certain factors. This study presents a very efficient prediction framework for assessing the quality of video streaming. The proposed system aims to precisely forecast the duration of play (total length of a particular movie), overall download speed, complete end-to-end delay, and beginning buffering latency using the extreme gradient boosting (XGBoost) machine learning algorithm. The results of the experiment show that the accuracy reached for playing time, end-to-end delay, buffering latency, and download rate is 0.985, 0.731, 0.937, and 0.927, respectively. This indicates that the system is effective in addressing the issue of extended waiting times during live broadcasting. Thus, the framework also offers a way to solve the live-stream delay to improve users' quality of experience. In conclusion, these findings are pivotal to the video streaming industry market to aid the service providers in designing their systems optimally, assigning the resources efficiently and increasing user retention.