MOS estimate improvement of ITU-T opinion model through erasure coding technique

摘要

Packet loss during video transmission can occur due to congestion at core network. The Packet loss can also occur when it get corrupted in the air interface due to the signal interferences, noise and low signal strength. Such erroneous packets are discarded by the lower layers and not pass to the application layer. In either case the sender information will not be reached to the receiver. The video frames are compressed before transmitting into network generally. Each frame is inter-dependent in compressed video; hence loss of video data packets can cause error propagation and results in video quality degradation and poor user experience at the receiver. This paper analyzes and assesses the video call quality of the received video, based on ITU-T Opinion model for video telephony and proposes improvement in the quality using the Application Layer forward Error Correction (AL-FEC) Raptor code. The original opinion model which is based on mere packet loss does not account for quality gain by erasure coding technique. We propose a method to improve the MOS metrics by recovering the packet losses through Raptor Code technique. The resultant quality score derived using proposed model has Pearson correlation coefficient of 0.91 and RMSE of 0.28, which are the better results than original model.