Optimizing HAS for 360-Degree Videos

In recent years, an increasing number of Internet-based applications have been released that use virtual reality for education, training, gaming, and various forms of entertainment. When transmitting an omnidirectional 360 degrees video over the Internet, it consumes considerably more data compared to traditional video streaming. To overcome the high network requirements and still reach a high Quality of Experience, HTTP adaptive streaming technology is considered for 360 degrees video streaming. In contrast to traditional streaming, the adaptation logic considers not only the current network conditions, but also the viewport of the user, i.e., which part of the 360 degrees sphere the user is currently focusing on. However, as the viewport of the user might change anytime, the adaptation logic is required to accurately predict the viewport in the next seconds of the playback to allow for an efficient and smooth streaming with a high visual quality. In this paper, we present novel linear programs that determine the optimal visual quality, which is reachable in a given network scenario using different approaches for viewport prediction. Our results are an important contribution for designing adaptation logics for 360 degrees video streaming, which allow for efficient data transmission in the network while reaching a high QoE in VR applications.