Marcus Weldon has two jobs: Corporate CTO of Nokia and president of Nokia Bell Labs. How does he balance these two roles? “I have a split personality,” he quips.
He could send the “Virtual Marcus” to do one of them, but then we’re getting ahead of ourselves. Suffice it to say that in all seriousness he interprets the CTO role as one in which he needs to look ahead and see the world as it should be. Bell Labs is where the technologies must be invented to help realize that vision.
Either way, his work needs to result in Nokia being in a leadership position, and as part of that, he needs to be thinking 10 years ahead of everyone else.
“My job is to see the future, because I might need 10 years to invent it, and that’s not because I’m lazy,” he said. It’s because some these things are very complicated—involving mathematics, physics, prototypes and a lot more to bring something to commercial status. A lot of it is timing, because if you’re too early in the game and develop a product that sits on the shelf, that isn’t earning anybody any money. Plus, the engineers will just have to re-do it anyway when the time comes. They also must be selective and pick the tough problems—they can’t work on everything."
He added, “The trick is to get to market just in time for when the specs are set and firm,” which is what Nokia Bell Labs did with things like Massive multi-input/multi-output (MIMO) and millimeter wave (mmWave). For mmWave, engineers had to work out the propagation characteristics and create Radio Frequency Integrated Circuits (RFICs) for multi-band mmWave with very small chipsets, which are now used in microwave gear.
Of course, the inventions at Bell Labs end up being shared with competitors like Ericsson—that's how the standards process works. However, "we tend to invent more things," he said, and then it's up to each competitor to add their own algorithms or "magic" to differentiate their products.
Bell Labs invented the mmWave chipsets that Nokia uses, the optical technologies that backhaul the radios, as well as its version of software-defined networking (SDN.) Bell Labs also works on emerging cloud technologies. “All those things we work on because we see the future end-to-end,” he said. “Then we build it,” and test it with real use cases at the Future X lab in Murray Hill, N.J.
Nokia CEO Rajeev Suri likes to talk about Nokia’s end-to-end (E2E) advantage in 5G and how the acquisition of Alcatel-Lucent enabled that. Unlike rival Samsung, however, Nokia doesn’t have a handset division to call its own; it sold that a long time ago. That means it has to build its own prototype devices—but it also reaps the benefits of that effort.
“We never lose value by not being in that business,” Weldon said, adding that Nokia has a lot of device patents even though it’s not in the handset business. That means it can make money by licensing the technology that it creates.
He acknowledged that Nokia has been winning more 5G deals recently, including more end-to-end deals that include things like backhaul, routing and optical, but it’s not clear if the growth in 5G wins is a result of the trouble Chinese vendors like Huawei are experiencing, or fully based on the merits of Nokia's own products. “We think it’s because we have leading technology,” he said.
So far, the questions about sourcing Nokia’s components—whether they’re coming from China—are more prevalent from government officials rather than customers. As for customers, “I think they view us as a trustworthy company,” he said. Even if one or two components are produced in China, they go through the same test validation procedures as any other components to ensure they’re reliable and secure.
In fact, Nokia today announced a new program, called DFSEC 2.0, focused on additional verification work in the areas of end-to-end (E2E) identity management, network slicing and SDN security and virtualization. Nokia also is opening a Future X Security (FXSec) Lab, built as an extension of Nokia's Future X network lab at Nokia Bell Labs, which will be open to communications service providers and industries to facilitate joint testing and verification of industrial automation solutions in private local area networks (LANs) and across public wide area networks (WANs).
What about NFV?
Operators in various parts of the world have grumbled to different degrees about how network functions virtualization (NFV) hasn’t lived up to expectations. NFV is premised on virtualization, which is something that had its genesis in the computer science community. Several years ago, telcos saw what was happening in the cloud industry and wanted to apply similar theories to telecom.
“We told them early on, that it’s not quite the same,” Weldon said, because telcos have some high performance and low latency requirements, and they’re not the same as the cloud industry requirements. “We always said, you will get some savings, but not as much as the cloud guys get,"he added. "I think what’s happened is reality set in.”
The savings operators are seeing are only maybe 20 percent or 30 percent and they were hoping for much larger numbers. There’s also more operational costs than they probably expected; operators have to manage their legacy networks at the same time they’re managing the virtual networks.
The real value of virtualization, he said, will come with 5G. That’s when traffic needs will be more sporadic. Today, traffic tends to just go up. In the 5G era, virtualization or cloudification will make sense because some things need to connect once in a while and others are “needier.” 5G is about diving more dynamically adaptive networks, which is not so much the case in today’s status quo networks.
Speaking of looking ahead, what about “Virtual Marcus”? It’s something the engineers at Bell Labs showed off to reporters and analysts during a tour of the Murray Hill, New Jersey, facilities after the Brooklyn 5G Summit in April. “Virtual Marcus” looks somewhat like a robot, but it’s really part of experiment the lab has been doing that has to do with the way people communicate in the future. After all, there must be a better communications paradigm than texting and emojis, right?
The way video calling is done today, two people have to be in locations where they’re happy to be seen, whether that means being dressed appropriately or the background looks acceptable. But, not everyone wants their facial expressions to be seen, or for the other person to see that they’re multi-tasking while they on the phone. The idea was to come up with a representation of a person that optimizes communications—so you’re sort of "there," but not really.
Bell Labs teamed up with a company called Ziva, which has extensive computer-generated imagery (CGI) experience in the movie industry. They liked the idea of transporting representations of people, and the “robot Marcus” was developed. It doesn’t sound exactly like how he talks, and Weldon said it’s only been partially successful so far because it’s turned out to be much harder to do than they thought. But, it’s meaningful in an augmented reality world and provides a subject of further research.