Perceptual quality assessment of HTTP adaptive video streaming

HTTP is rapidly evolving as the most popular technology for real time multimedia content delivery over the Internet. The 'best-effort' nature of Internet means that video streaming applications will need to overcome rapid bandwidth fluctuations and varying resources availability. In order to prevent the client's buffer under-flow, maximise the playback quality and improve the user quality of experience, rate adaptation is performed over HTTP streaming. This paper studies the Dynamic Adaptive streaming over HTTP (DASH) and investigates an adaptation logic scheme based on a cost function of the throughput estimation and the previous quality index. Experimental results over a test-bed platform indicate that the selection of the cost function weights, may significantly affect the network utilization during video streaming. In addition, subjective evaluations of the perceived video quality based on MOS scores produced by the absolute category scale method, suggest that the perceptual video quality is similar in the two cases where the adaptation logic aims at either keeping the transmission throughput constant, or the quality index close to the previously index used.