MobiledgeX Edge-Cloud with Google Cloud’s Anthos: A Powerful Platform for 5G, CDN, and Privacy-Centric AI Applications

April 29th, 2021

Headshot for Sunay Tripathi

Sunay Tripathi

Chief Technology Officer & EVP of Engineering & Product

Public cloud providers are moving into the core of mobile networks with their edge infrastructure offerings. That is a good thing since they offer a highly optimized and valuable alternative of supply. However, being able to control how and when they are used is very important.  MobiledgeX’s Edge-Cloud with Anthos enables that control and we just announced a beta program for telecom operators who are interested in trying it out. This combined stack approach gives operators the freedom and flexibility to orchestrate and automate select 5G applications, for example mission-critical AI applications that process video feeds from business and enterprises, that demand a higher level of security and privacy. 

You might wonder what is different about this stack compared to a pure public cloud stack that you might already use? Let me add some context and detail before I explain more of the technical benefits that MobiledgeX Edge-Cloud on Anthos can bring to telecom operators.

5G Distributed UPF Applications on Edge 

5G standards allow various applications (like evolved packet core) to be separated into a control plane and a data plane. One of the primary goals of this separation is to allow data packets to be routed closer to the location of the user while running the control plane separately to make it easy to centrally manage these functions. 

Figure 1: An example of disruption of home routing while roaming, with a mobile device user traveling from Europe to the United States.

For a telecom operator, routing for roaming customers/subscribers is a prime example of separating the data and control planes. Operators want to control the admission of their subscribers onto their network (admission control) more centrally because authorization, authentication, and accounting procedures are still centralized in core clusters, yet operators don't need to route packets to anchor based on the subscribers area code to verify if you are real (and if you have paid your bills). The easy fix with 5G service-based architecture (5G SBA) is to push the user plane function (UPF) for routing close to the roaming subscriber, and if subscribers are in different regions or countries, the point of presence in that region becomes the best platform for deploying UPF. 

This is a use case that we are working on with several of our telecom operator and developer partners where we can take the UPF to different places based on the presence of subscribers and give it more (or less) CPU/GPU based on the number of subscribers at a given location. We encode this in the MobiledgeX Edge-Cloud automation policy for UPF via “Autoprov” and “AutoScale” policies. We then manage the software including the versions, security, and dependent services which include load balancers, public/private IP addressing, and other networking needs under the MobiledgeX Edge-Cloud “Autonomous Deployment” engine. 

The MobiledgeX Edge-Cloud platform brings full automation and orchestration services to the distributed UPF so that subscribers can benefit from 5G services (faster speed, lower latency, higher bandwidth) while operators don’t have to raise subscriber fees because of the efficiencies MobiledgeX brings to their network services. 5G SBA is not just an incremental signaling standard, but more of a generational architectural shift of distributed computing to enable these new services. 

CDN and AI Applications on Edge

What happens when the same roaming subscriber wants to watch video content or utilize an AI service which does video processing from a mobile device? Logic says if we are routing his data traffic close to the subscriber, then we should allow the video processing or caching to happen close to  the subscriber as well. 

The reality is that some applications -- like those used for CDN, AI, and video processing -- are often widely distributed and have a fundamental requirement for fixed or bounded latency, not just low latency. These applications also have a greater need for reliability, security, and privacy elements since those applications are running in businesses, enterprises, and public spaces that are far removed from a centralized data center. Bringing all this traffic to a remote location isn’t technically feasible and is financially expensive for many of the reasons outlined above.  

These are the kind of applications that can benefit from the MobiledgeX Edge-Cloud orchestration and automation engines. I used roaming as an example, but even within the operator network in the region where those applications are being used we need to push these services out closer to where subscribers use them for the same reason.  With all of these data and network elements getting so highly distributed, the demands for latency reduction, security, and privacy are helping generate the excitement about the benefits that edge computing can bring.

MobiledgeX Edge-Cloud to Deploy 5G UPF, CDN and AI Applications Serving Mobile Subscribers

It should be apparent by now that the MobiledgeX Edge-Cloud platform provides application owners with orchestration and policy-based deployment controls and tools. To support the needs of operators, our software platform ensures that control over deployment policies, life-cycle management, resource allocation, monitoring/billing, etc., stays under the control of the operator OPS team via the MobiledgeX Edge-Cloud operator console and our controllers run in the operator network itself to support its privacy and data sovereignty needs. This is the driving force behind MobiledgeX’s “Private and Trusted” cloudlets, which is different from the “Public Edge” which we operate with Deutsche Telekom and other European partners where the developer deploys and manages his own application via the MobiledgeX Edge-Cloud developer console inside the operator network. These elements need to work the same way with the MobiledgeX Edge-Cloud platform over their existing OpenStack and VMware-based infrastructure as well as when they get deployed outside their network for roaming subscribers.

MobiledgeX Edge-Cloud with Anthos

Telecom operators don’t have infrastructure in many of the places where public clouds are present. When Google announced that Anthos now works with the MobiledgeX Edge-Cloud platform to allow control of the policies and application deployment for operators it became very natural for us to work together to support the combined stack as a way for operators to have one more tool in their arsenal. It also opens up even more possibilities in the future where we can simulate the cloudlets in public clouds so that operators and developers can work together in a public edge environment to qualify and certify applications. The operator operations team now can actually move applications for deployment on a private and trusted edge, which can be based on Anthos -- in bare-metal form -- as an open, on-premise offering that supports the same infrastructure and the same Kubernetes container management environment used in public clouds. 

Figure 2: UPF Pipeline: Automating 5G UPF from test to production using Github action.

Learn More About MobiledgeX on Anthos

This beta program underscores the ongoing work that MobiledgeX has done with Google Cloud and the Anthos platform, which began when MobiledgeX was named as a horizontal solution provider for Anthos for Telecom in December 2020. If you’re an operator interested in participating in the MobiledgeX Edge-Cloud Beta Program on Anthos, please contact Ulf Andersson, MobiledgeX Senior Vice President of Global Business Development and Sales ([email protected]).