Content for  TR 22.856  Word version:  19.2.0

The mobile metaverse allows users to access an endless virtual world at anytime and anywhere through their terminals. The mobile metaverse are expected to behave as the real world, which means in addition to rendering a virtual environment like the physical world, the perceived spatial-temporal consistency is also the key point to achieve an immersive location agnostic service experience. In mobile metaverse, spatial-temporal consistency for single user could mean, for example, dropping a virtual pen and seeing this pen fall subsequently. While for multiple players, this consistency could mean, for example, that one person cuts down a tree and other people see the tree falling down. This user experience requires the motion-to-photon latency in the range of 7 ms to 15ms [5] at least for a single user viewing the consequence of her own actions. Immersive VR requires the delivery of massive amount of data (in the order of Gigabyte) at ultra-low latency (less than 20 ms) [54]. It should be noted that the computation resources for rendering involved in the mobile metaverse is different from the cloud gaming and traditional VR. For example, running a typical massively multiplayer online game today requires multiple tera FLOPS of graphics horsepower, and the demand is expected to grow by two orders of magnitude to create fully immersive mobile metaverse experiences. [55] For the mobile metaverse world, distributed computation is an inevitable processing mode, so the selection of proper servers and data centers should consider the requirements of network delay, processing delay, storage and computation resource. The goal is to minimize the user's perception of delay. Therefore, in order to obtain consistent experience in mobile metaverse service anytime and anywhere, deep collaboration between mobile metaverse and 5G network is needed. The potential collaboration aspects may include caching location, computation location, communication path, traffic scheduling and resource allocation in network. For example, when a service request emerges, the network control policy needs to coordinate the selection of (i) caching locations to provide digital objects, (ii) computation locations to execute service functions, and (iii) communication paths to route all associated data streams, jointly optimized with dynamic decisions on (iv) traffic scheduling and (v) resource allocation at all network locations. [55]

There are about 12,000 players sign up for a popular game and appear simultaneously in a specific setting, such as Eve Online in 2021. Due to the limitations of the existing server processing, it is not possible to support such a large number of high concurrency, so the network and application server need to cooperate to support the distribution of visitors to other servers while ensuring low latency requirement by XR applications.

1. Bob is a player who attends a popular AR interactive game and gathers with others in a shared environment, they are aware of each other's action so that they need high synchronization.
2. The service provider will provide deployment information of each server to the 5G network, and request the Bob's physical location and transmission delay in 5G network.
3. According to the cooperation agreement with application, 5G network will expose information to service providers, including the physical location and network delay of specific terminals or a group of terminals. The network delay includes the delay inside 5G system (UE to PSA UPF) and the latency information between PSA UPF and some potential servers.
4. The new server is selected by the service provider according to the UE location, network delay, business requirements, computation resource and storage resource of application servers. The decision result will be sent back to 5G network. Then the 5G network can then formulate corresponding policies for the service flows.
5. The content information will be synchronized to the new server in real time. 5G network should support the ultra-low latency data transmission, potentially among multiple operators.

Bob will have a good experience in this interactive AR game.

3GPP started the work of edge computing from R15 to R18. In R15, AF influence mechanism is introduced to inform the 5G network of the application deployment information to assist UPF selection. In R16, 5G system supports QoS monitoring mechanism for end-to-end delay monitoring for URLLC services. In R17, 5GS supports to solve the problem of edge DNS selection and service migration between different edge platforms. In R18, the work focuses on the edge computing platform access from other operator network, and the distribution of network policies for a group of local UEs.

No potential new requirements have been identified.