Impairment-factor-based audio-visual quality model for IPTV

This paper presents an audio-visual quality model for IPTV services. The model estimates the quality of high definition (HD, 1920times1080 pixels) audio-video as perceived by the user. It is developed for being applied to network planning and packet layer quality monitoring. Addressed degradations are mainly audio and video compression artifacts and packet loss. Results demonstrate a mutual influence of the perceived audio and video qualities, and the predominance of the video quality for the overall audio-visual quality. The model is based on impairment factors which express the quality impact of the different degradations. Those factors are computed from parameters extracted from the bit-stream or packet headers, such as audio and video bit-rates and packet-loss-rates, making the model computationally efficient. The model correlates at 97% with unknown subjective ratings. For comparison, we have developed a more classical audio-visual quality model based on audio and video qualities and their interaction, which shows lower performance than the impairment-based approach. Also provided is a comparison with models described in the the literature.