August 3, 2018
Next-generation communication services will require a large diversity of service-quality requirements:
Higher end-user data rates and Higher Cell level throughput to meet enhanced mobile broadband (eMBB) requirement
- Low Latency ( RRLLC)
- Massive number of devices for machine type of communication (eMTC)
5G networks need to support multiple QoS for a variety of applications/services, as opposed to current networks that are designed for a limited set of QoSs. It will also have to support multiple levels of various level of QoS ( e.g. premium) among various tenants on the network. Depending upon use cases, there will be requests for simultaneously meeting mutually exclusive or overlapping requirements from different tenants. They will require guaranteed performance levels and isolation from each other. Additionally, tenant’s end customer’s service requirements will vary over time and network servicing area.
Traditional CNs, designed for purpose-specific service requirements may meet some of these requirements. However, diversity of requirements for service quality, time, and region can be met only by a more flexible, scalable, and agile network architecture.
Network slicing (NS) can support this in an efficient manner. In NS based design, virtual network functions (VNF) run over common infrastructure comprising of a resource pool of physical resources. Network slice management and orchestration instantiates, manage, and terminate E2E (logical) network slice instances on demand, by tailoring NS templates and allocating necessary virtual and physical resources and functions from the pool.
An E2E network slice is defined from the device/terminal to the IP interface of a different network and includes RAN, transport and CN along with operational and business support system functions. Network slice management and orchestration include administration of multi-vendor, multi-technology, and multi-operator scenarios, including roaming and SLA cooperation. Network functions are placed either in the edge cloud or the centralized/remote cloud. E2E quality of experience can be brought in by bringing in subscriber and application awareness to several functional areas in the network instead of just being in the CN. The service quality negotiated in SLAs needs to be guaranteed and fulfilled throughout the operational lifetime of the network slice instance. Powerful predictive and semantic analytics, including machine learning and AI, will help to assure the service quality for all concurrently operating network slices.
A network slice management and orchestration layer can support QoS management by evaluating the current network slices performance and orchestrating necessary resource changes in VNFs in different RAN, CN, and transport to meet the required QoS. For making E2E negotiation of QoS effective, application and service awareness has to be built at various network points of presence.
Reference IEEE 5G and beyond Technology Roadmap White Paper