End to End Video Quality Prediction for HEVC Video Streaming over Packet Networks

Advances in multimedia devices, video compression efficiency and increased availability of network bandwidth have increased the proliferation of multimedia applications. However, the proliferation of these applications has also increased the need to maximize available network resources and also meet end users’ Quality of Experience (QoE) and expectations. We present results for end-to-end video quality prediction based on the combined impact of application and network related parameters on Highly Efficient Video Codec (HEVC) based video streaming. The test covers 360 scenarios and several distinctive dimensions such as video content type (CT), quantization parameter (QP), frame rate and network packet loss. The results show that, video quality depends on the content type, initial encoding parameter setting such as QP and bitrate. A lower QP results in a higher initial video quality and vice versa. However, an initial higher video quality requires a higher video bitrate and thus higher transmission bandwidth which could incur more packet loss and quality impairment in a bandwidth limited network. Additionally, the results show that packet loss degrades video quality with a higher loss rate resulting in greater video quality degradation. Based on the objective test results obtained, we performed a thorough statistical analysis to understand the combined influence of these dimensions on QoE as perceived by users. We believe that such study could extend existing objective methods of measuring and predicting acceptable user perceived quality especially for the newly released HEVC video coding standard. KeywordsQoE; HEVC; QoS; Video quality evaluation; video