Modeling of Edge Server Cache-Reservation for Virtual Reality Applications
Ishfaq Bashir Sofi, Rodolfo W. L. Coutinho, M. Reza Soleymani · 2022
Envisioned virtual reality (VR) systems include mobility as one of the important features of users' when interacting with virtual environments and objects. Thus, VR content encoding will be processed on cloud servers and delivered to users through the network, as head-mounted headsets (HMDs) have limited computation capabilities. Therefore, network bandwidth becomes the bottleneck for cloud-aided VR applications. In this paper, we proposed a hierarchical cloud/edge VR content cache architecture and a cache reservation policy for the cache of highly popular VR content aimed at reducing the latency for content delivery to VR users. The proposed mathematical framework models the different network conditions experienced by the users, 360° video stream popularities, and a proposed cache reservation mechanism that allocates different amounts of cache memory at edge servers to cache VR content encoded at different resolutions. Obtained results show that the latency for VR content delivery is sensitive to the portion of cache memory reserved for each VR content encoding quality as well as the popularity of the requested content.