Mobile operators can’t transform the business of telecom simply by offering faster speeds and new services.
The magic comes from making significant advancements in areas like network site planning, building and maintenance, and careful management of the field force executing rollout. Much like how Amazon’s online shopping business is about industrialized logistics and supply, telecom’s business is the same but promising customers connectivity rather than shopping.
1&1 in Germany knows this well. The greenfield operator has just launched its fully virtualized 5G network, successfully achieving hardware and software industrialization, establishing a common data set and implementing AI-driven automation. With dozens of vendors, systems integrators and partners all contributing to this ambitious project, it is no small feat to have established an advanced network operation powered by a common data set.
This is a milestone every telco will need to reach, whether existing or greenfield.
As Igal Elbaz, SVP, Network CTO, AT&T, commented recently, “We keep talking about open and cloud RAN, the excitement of the architecture – but real life is building the network. For a network of our size, nationwide across north America, that takes a lot of effort and planning.”
At Mobile World Congress Barcelona, Network Media Group hosted a panel with some of 1&1’s key technology providers to explore how the operator has been able to streamline one of telecom’s most ambitious network rollouts and what the rest of the industry can learn from the success achieved to date. Joining me were Mory Lin, VP of IoT/Embedded and Edge Computing at Supermicro and Caroline Chan, VP of the Network and Edge Group at Intel.
The conversation anchored on the reinvention happening within telecom and the evolution that has occurred since Rakuten Mobile built the industry’s first fully virtualized network in Japan. 1&1 is benefiting from Rakuten’s years of experience operating this network, along with the latest advancements in technologies, network designs and strategies.
Even the language has changed. I contrasted the difference between how networks are described today versus previous generations. Builders of 2G, 3G or even 4G networks would refer to points of presence using terms like cell sites, base stations and mobile switching centers. At 1&1 network, these network locations are now central data centers connecting to regional data centers connecting to edge data centers. This has obsoleted much of the vernacular that has been so common within telecom going back decades.
It is Germany’s fourth mobile network and the first fully software driven network in all of Europe.
How did 1&1 deploy its network so quickly? By completely embracing industrialization of the network. Building a complex network involves a slate of unsexy processes that must be repeated at tens to hundreds to thousands of sites. Within each site are just as many components, all managed and deployed by the same number of people working in parallel.
1&1’s network is powered by automation enabled by a common data set available across all systems throughout the network.
Industrialization at this scale requires a new hardware supply chain. In the 1&1 network, all locations can choose from a hardware template of Intel-powered Supermicro servers based on the size of the site and functionality deployed there.
Similarly, a standardized software supply chain based on cloud-native microservices was implemented. This software can be remotely deployed, maintained and managed throughout its lifecycle.
The panel discussed how AI can improve OSS/BSS functionality and flexibility given the quantity and availability of telemetry and other data that powers proper network operation. AI at the edge is an important aspect of this because it benefits from lower latency for faster decisions. The importance of edge AI led one presenter to suggest it will be running the network within a year. The common data set provides the data foundation for AI to have network-wide impacts on managing the network.
Truth in AI is important because AI is only as good as the truth (i.e., data) it’s given. In the real world, when people have different opinions, they can come to different conclusions. That doesn’t work with AI because it is not acceptable for a machine to have an opinion on data that is incorrect.
This brought the conversation back to the importance of extending network industrialization to the entire software and data supply chain.
Despite its “greenfield” label, 1&1 stands as an example of telecom reinvented that existing operators can learn from as they plot strategies of their own.
