5G Core - Above and beyond the core

Nickname stick to the people, and the most ridiculous are the most adhesive, somewhere I read this exciting line by Haliburton. Evolved packet core (vEPC) transforms into 5G core. I started my career architecting mobile switching center (MSC/VLR), the predecessor of vEPC. While all the associated network functions may be far-away from core networks in the carrier infrastructure landscape, the industry calls it 5G core. We are naming it core, maybe because it anchors the multi-access sessions! Probably, by 6G or 3GPP 17 release, the segregation goes away aligning with distributed cloudification. Inherently, we never stop segmenting the larger applications into smaller ones based on some rules for ease of understanding, designing, and implementing. 5G core is no different; we still have grouped network functions based on user plane, packet controller, network resources, signaling, and so on.

Intending to align the protocol with the cloud world such as JSON and HTTP2 as well as services repository, 5G core service-based architecture aims at converging services to one network function. Earlier the non-orthogonality burdens our core architect to implement one function in multiple nodes. But that brings an interesting dilemma and an opportunity to deploy one network function at different locations depending on use cases and quality of services.

It's one thing to build a disaggregated telco networks functions and its another to meet service providers expectation to exceed the scale, performance, cost of ownership, and ease of deployment aspirations. Moreover, new revenue-generating use cases and enterprise digital transformation opens up a new market for 5G core. The nimble 5G core vendors might be closer to an exact cloud-native implementation, smaller footprint, and Kubernetes based orchestration that are key to meet these variable industry demands. The very first time in the telecom industry, the focus is more on enterprise use cases. Disaggregated 5G core design and deployment models are the main reasons for these new opportunities.  

Service providers may continue to deploy 5G core in private clouds on RedHat or VMWare. There are valid reasons to do so as of now. But enterprises don't have to follow the same deployment model. User plane functions (UPF), session management functions (SMF), or even the access and mobility management function (AMF) could become part of the edge in the private cloud such as edge data center or regional data center. The rest of the network functions could become part of a public cloud. Cost of ownership, data privacy, performance, and ease of network operation may drive the different hybrid cloud deployment. It would be interesting to see several quantitative modeling as a tool to solve this problem based on policies.

Traditional virtual evolved packet core (vEPC) has been playing as one of the elements for driving a private cloud network today. One of the tier-1 telecom vendor's private network positioning statement is "don't wait for 5G". But, to address the variable scale, to have multi-tenant network function, and to minimize 5G core functions deployment at the edge for most of the cases, we need 5G core network functions. Large enterprise deployment may justify small footprint vEPC. The business case for deploying in medium and small-sized enterprises could be more persuasive with 5G core. Cloud service providers and 5G core vendors are trying to speed up the private network deployment with use cases focused model agonistic to underlined private network infrastructure. That is where cloud-native, low cost and small footprint 5G could play a critical role.

Wireless broadband is another use case in which 5G core delivers the service with better cost and better performance. To compete with wireline broadband service providers, the wireless service provider need to be sensitive about resource cost and speed. Along with disaggregated and thinner 5G edge network functions, it is the uniquely implemented user plane functions that would make one 5G core different from other vendors. Multi-access edge (MEC) and network slicing selection function are instrumental in delivering such service meeting specific SLA. 

In my opinion, UPF, NSSF, SMF, NRF, and AMF are the five network functions that differentiate one 5G core vendor from others. The rest of the 5G core network functions will be a function of these five network functions. The user plane function (UPF) enables hundreds of use cases for industrial scenarios. UPF addresses the service provider requirements of edge analytic, enterprise monetization, and latency. Flawless implementation of application detection and QoS management for each session in UPF is the dream project for most of the telecom architects. UPF to be scanned carefully for ultra-reliable low latency use cases (uRLLC) in 5G core outside the RAN. Accelerating the UPF functions by offloading some of the high CPU intensive tasks on Intel SmartNIC card may further help the service providers' needs. Network slicing selection function (NSSF) is a few of the 5G components that are going to generate new revenue for service providers. NSSF will also enable innovation in delivering some of the use cases from carriers to enterprises. 5G core vendor would differentiate itself with the ease of creating innovative use cases collaborating with others orchestrators and 5G RAN and MEC. Session management function (SMF) defines how scalable and efficient 5G core is in meeting service provider performance and security needs. 5G core vendor allocates the best of the best architects among their team for SMF. Seamless user plane function (UPF) interface management, mercurial IP address management, and intelligent traffic steering determines the smart session management, specifically for massive machine type communication (mMTC). Network repository function (NRF) is very matured in the IT and cloud world. Network functions profile management and network service discovery features are part of NRF. NRF is one of the components in addressing service providers' need for network infrastructure integrity and latency in the 5G network. Infrastructure sanity management is possible by integrating AI/ML-based solutions with a resilient NRF.   

Hopefully, all the standard and compliance communities would very soon give similar attention to ease-of-integration as it delivered to the network function specifications. Deployment and integration sometimes become so critical for service providers that it sometimes forces the service providers to ignore best performing 5G core vendors over the best deployable 5G core solution. Ease of integrability with third-party solutions is positively co-related with ease deployment and integration.  Orchestrator, assurance, self-organizing network (SON), and CI-CD-CT tools are the areas most R & D investments are happening now, during PoC and pilot projects from 5G vendors. Traditionally, telecom vendors focus on addressing deployment and integration problems when the product is ready to be introduced. Established and matured 5G core vendors have partially integrated the deployment specifications since the product inception. Deployment and integration plug-ins culturally are considered as secondary areas of focus for the new players and the startups.

With the cloud service providers aligning their enterprise growth strategy with 5G core, market analysts might have gone to the whiteboard to adjust the total addressable market and to redraw the diffusion of innovations curve. Admittedly, we are all in the race among multiple 5G solutions, products, use cases, and service players to capture our pie. Sometimes competing with some and sometimes complementing with some. Connect with me for exploring the complementary synergy in driving 5G core solutions, services, and use-cases growth.