No-Reference Modeling For Quality Of Experience Evaluation On IPTV

With the high speed of IPTV development, IPTV services are becoming one of the most promising applications over the next generation network. End user’s subjective perception on IPTV services, namely Quality of Experience (QoE), is increasingly gaining attention by service provider. In this paper, a no-reference QoE evaluation model is proposed to forecast and improve the end user’s perception on IPTV services. The specific works are shown as follows:Build a simulation environment of IPTV services, specifically setting up an IPTV server to establish an experimental environment to simulate the WAN environment and publishing a video on-demand player which can simultaneously record the video and extract parameters.Conduct a subjective video quality assessment experiment using the Single Stimulus Continuous Quality Evaluation (SSCQE) method to collect Mean Opinion Score (MOS) score. After the SSCQE process, all evaluation votes are through a strict screening process according to ITU-R P.1788to achieve effective votes to calculate the MOS of each video.In this pa-per, we focused on the network video impairments and some video Key Performance Indicators (KPIs). Through our research, it was summarized that some KPIs can affect the degree of video quality robustness, which means that even under same network environment, different IPTV video sequences may have various users’perceptions. Our proposed model derived a mapping function to assess IPTV subjective quality from video bit rate and network-related quality of service parameters, amended by some video KPIs which affect the degree of video quality robustness due to network impairments. The Pearson Correlation Coefficient between simulated QoE score and subjective quality score MOS is achieved0.957and the Root Mean Square Error is0.248. Therefore, our proposed model can be applied to evaluate end user’s perception on IPTV services with considerable accuracy.