Evaluation of Video Streaming Quality Degradation in Lossy Environment

The article describes the issue of assessing the impact of packet loss in the network on the degradation of video quality received by users. The case of live streaming service was considered here, where potential error correction by retransmitting lost packets is not implemented. The authors described the results of research conducted in a test laboratory environment under full control of network traffic. The prepared set of reference video samples in three resolutions was encoded using H.264 and H.265 codecs. The video samples were transmitted over the network with different levels of packet loss and recorded at the receiving end. The degraded video was assessed using objective and subjective methods. The obtained results were used to build a machine learning model that can be used to predict the video quality experienced by users based on the designated objective metrics. The authors thus indicate that the machine learning model allows for obtaining better video quality prediction results than the previous modeling based on single objective metrics.