An Open Framework for the 5G UPF
There are many ways to deploy 5G
But there is one way to avoid vendor lock in - minimize latency - reduce complexity
There are many ways to deploy a 5G network - but most of them will cost you dearly.
The 5G Core introduces huge opportunity for Mobile Operators to leverage the throughput and latency improvements that 5G New Radio provides in a flexible and standardized multi-vendor environment.
Isn't that all good news?
3GPP defines the UPF as a single element decoupled from the control plane that handles all aspects of data plane packet processing between the RAN and the DN through one or multiple instances. 3GPP concentrates on the access control, session management and packet processing aspects. In other words, 3GPP functionally prescribes the roles played by the serving/packet gateway in the EPC.
Operators, however, demand a pragmatic implementation for the UPF that offers enough flexibility to accommodate multiple use cases and regulatory obligations. These usually require niche user plane services and applications that, due to latency, performance and efficiency reasons, should ideally be part of the UPF ecosystem in the 5G Core.
Why is that a problem?
The problem with this approach is that the standards do not address how network services, which traditionally resided in the GiLAN, must be architected and deployed to leverage 5G throughput and latency improvements.
The 3GPP network architecture for the user plane groups and isolates functions into UPF entities. When introducing services in the core, operators might have to choose between
- A single anchor UPF that integrates applications. A closed and proprietary UPF element imposes functional and business restrictions to Operators, introducing vendor lock-in. The Operator will become dependent on the UPF vendor for future services and use cases.
- Deploying a collection of UPF elements from different vendors. Although the 3GPP architecture for the user plane provides with both control and user plane interfaces to accommodate this model, this approach will inevitably add processing latency and complexity to the service rollout process and its scalability.
- Deploying services in the N6 interface, as a continuation of the traditional GiLAN model commonly adopted for the EPC. Again, this model will not allow the Operator to benefit from efficient packet processing, simplified orchestration and scalability, introducing multiple hops in the user plane.
Solution to the problem posed by the 5G UPF
At Openwave Mobility, we treat the 5G UPF as an open framework that acts as the only user plane data processing element. However, as opposed to a monolithic element, as suggested by 3GPP, it leverages service decomposition to create an extensible framework through self-contained stateless functions. This architecture for the UPF, as an end-to-end data plane framework, is based upon the following key design principles:
Openwave Mobility’s design principles for the 5G Core User Plane Function
UNIFIED DATA LAYER
LEGACY CORE SUPPORT
Actionable figure: please click each box to understand the design principles and the accompanying benefit
Openwave Mobility offers multiple deployment models for the UPF, whether as an Anchor UPF as well as a Service Function UPF. For more information about Openwave Mobility solutions for 5G, please contact an Openwave Mobility Representative or fill up the details in the contact form.