Table 2. Network data rate and latency requirements for 360° Video Cloud VR An IaaS for AR/VR Cloud services must meet specific requirements for each use case. These requirements include asynchronous terminal cloud rendering, which can reduce latency requirements for cloud rendering and network transmission. However, black border, image quality degradation, and other issues that hinder usability occur when rendering and streaming latency are not ideal. According to industry test results, latency is expected to be less than 70 ms. Currently, China Mobile`s asynchronous rendering technology is being applied in its commercial trial of virtual reality in the cloud. Connectivity: Most of the best AR/VR services are enriched by 5G networks. 360° video streaming will increase the need for bandwidth, and services that extend video streams with mobile capabilities hosted by the edge require very low round-trip latencies. The exact location will be required for many services such as tourism, gaming and outdoor sports, while the network slicing provides the flexibility to meet the elastic capacity requirements of augmented events.
To keep end-to-end latency as low as possible, codecs can support slice-by-slice encoding and decoding, as well as slice-by-slice transmission. This makes it possible to perform several operations simultaneously. Also, when selecting coding structures, we recommend that you do not select multiple frames or frames back. Data frames should appear immediately after decoding. Since physical hardware requirements depend on the use cases actually served, efforts are underway to standardize hardware layout, particularly the Open Compute Project (OCP)[13] and Project19[14], which aim to define a common cross-industry server form factor and create a flexible and cost-effective data center and edge solution for operators of all sizes. The general service architecture for Cloud AR/VR is shown in the following figure. The COMPUTing aspects of the AR/VR experience are handled in the Cloud Edge, with the visual experience being made available to the user terminal via a 5G network connection with low latency, high capacity and guaranteed throughput. The same 5G network connection also carries messages from the terminal`s input sensor to application processing. There are a number of aspects to consider when it comes to VR terminals. These include the state-of-the-art data throughput of integrated 5G communication modules, 5G frequency bands, IP protocol stacks, and requirements for virtual reality and 5G integration.
Different cloud AR/VR services require the network to be able to easily load and manage different services and applications. Various AR/VR cloud services and applications are deployed on different network sites with edge computing as needed. Total network latency is directly related to the length of a signal path and the number of intermediate routing/switching elements in this hotfix. The following figure shows the impact of end-to-end latency as network-side features move closer to the edge of the core network. VR/AR are here to stay. Technology will mature rapidly as companies explore new ways to leverage technology and improve revenue streams (especially in marketing/retail). The next generation of networked telecommunications infrastructure will reduce the barrier to entry in costs and broad adoption will follow shortly. Mobile edge computing will bring wireless augmented reality to the masses and dramatically change lives.
The long latency observed in the H.26x series of video codecs is mainly due to the following reasons. Current virtual reality applications are just over 360º video/panoramas. While it can be incredibly detailed, low-quality 360° video requires at least a 30Mbps connection, with HD quality streams easily exceeding 100Mbps and retina quality (4k+) reaching an area close to Gbps. All of this means that virtual reality deployments in 2017 will be largely limited to device location (mainly for low latency reasons) while waiting for operators and ISPs to update their network infrastructure. Currently, 4K panoramic video only requires a data rate of 20 to 40 Mbps and a latency of 50 ms. As 5G networks contribute to a significant improvement in data rate (over 100 Mbps) and latency (less than 10ms), users will enjoy a more comfortable viewing experience. Thin client functionality in the endpoint provides the minimum functionality needed to present the VR/AR experience, with all additional functionality being moved to the cloud. Since significant amounts of data are always exchanged between cloud-based processing functions and the user`s end device, the use of low-latency coding techniques is necessary, even if they are associated with less than optimal compression factors. The cost and complexity of IT systems in all areas often followed a cycle in which early solutions required customized technology and significant resources to overcome less optimized algorithms, while later services benefited from reusable general-purpose platforms. The adoption of cloud technologies across the industry is strengthening the pay-as-you-go model, which provides access to previously inaccessible layers of processing that can now be used to deliver immersive experiences.
Low-latency, high-speed networks provide instant access to remotely stored data while providing a local computing experience with a data center-based system. Figure 10. Serial processing: MTP latency vulnerable to more than 20 ms *Note: The latency in the table indicates the Round Trip Time (RTT) latency requirement on the radio access network (RAN) side. Rendering and encoding/decoding latency is not included. VR terminals integrate 5G communication modules mainly in two ways. 5G chip modules are used directly for integration, which places high demands on the design of VR terminals. The other is the combination of 5G communication modules via plug-in interfaces (min? PCle or M.2) or via vibration mounting (LCC or LGA). VR terminals can select one of these modules based on specific product requirements.
The most important factor in cloud AR/VR solutions is the transfer of processing capacity from the local computer to the cloud. A high-capacity, low-latency broadband network enables responsive interactive feedback, cloud-based perception, real-time rendering, and delivery of display content. Cloud AR/VR is a new type of AR/VR that has the potential to change the current business model and expand operators` cloud platform services. A distributed IaaS can serve different users: internal services (implemented as virtual network functions – VNFs), higher-level internal services (e.g. carrier-owned AR/VR services) as well as external services. Depending on the type of hosted services, different requirements apply, as described below. Enabled by network slicing technology, shared services and virtual reality service flows in the cloud run on different slice channels. When network congestion occurs, the bandwidth and latency conditions of VR services in the cloud are still intact, reducing the negative impact on the user experience. The bandwidth and low latency requirements for these heavy video technologies are at a level unmatched by any other current technology, and our current networks are extremely ill-equipped to handle even VR streams in HD quality, let alone immersive 4k+ virtual reality.
The VR cloud system is very sensitive to transmission latency. This requires not only high throughput, but also low end-to-end latency. 5G networks with high data rate and low latency are suitable for transmitting VR services in the cloud.