Pokémon GO first to DT “Edge network in the core”

Deutsche Telekom is building an “Edge infrastructure through decentralized cloudlets.” Niantic Pokémon GO will be one of the first apps, with other AR/VR nets joining in. Controlling autonomous cars probably cannot work, but coordinating them and working with the connected car’s onboard computers could yield remarkable results.

Cloud servers with ample processing power will be located in one of Deutsche Telecom’s 17 facilities that connect all of Germany or one hop closer to the consumer. 

What I’m calling an “Edge/core network” does not deliver the 1 ms to 10 ms Nokia CTO Marcus Weldon once thought would be required for augmented reality. The DT network is designed for 20-25 milliseconds away, in parts of DT’s backhaul system closer to the cloud than the consumer.

Imagine that a different player controls each of the figures in the illustration, moving them around and interacting with each other. “Niantic’s Real World platform for Augmented Reality”  supports “real-time mass multiplayer experiences, persistence and visual occlusion, through contextual computer vision.” Watch the two-minute video below to understand what that means, hard to describe in words.

5G enthusiasts assume the “Edge” is close to the consumer: last mile, C-RAN, or maybe metro pop. That’s what the telco network people assume. I thought the same until Jason Hoffman of MobiledgeX described the distributed cloud due in 2019 his parent company, Deutsche Telekom is building.

MobiledgeX offers an API that allows developers like Niantic to do their thing. It’s designed to connect other carriers as well, perhaps Telefonica Deutschland. I would have expected MobiledgeX’s parent company to baulk at that. Hoffman thinks that would be like a phone network that couldn’t connect to others. A Verizon user wouldn’t be able to call a friend on AT&T. A DT exclusive system would be much less attractive to prospective customers.

This may be an “Edge network for the early years,” although it doesn’t have the many thousands or tens of thousand distributed cloud servers required for even lower latency. No carrier is committed to 1 ms latency, the dream of 5G planners. Verizon and AT&T are building to 10 ms. Swisscom’s early tests are 25 ms. Neither they nor anyone else have publicly committed to building a large and expensive network closer than DT’s. 

“Edge/core network” is about as logical as “jumbo shrimp.”  I wanted to be more precise because a dozen different definitions of “edge networks” is confusing people, myself included. 

A logical next step is interchanging traffic information. DT and experts believe connected cars can never be controlled from the cloud, even one closer than 20-25 seconds. What would the car do when it was out-of-range or in an area with a malfunctioning cell? What would happen if a hurricane disabled part of the network? The car engineers don’t believe solutions are likely anytime soon.

But “edge/core networks” can work closely with the computers in the car, providing a constant feed of crucial information. The interaction can be informed by the information from other cars, sometimes as simple as which lane to choose to even the flow at toll booths. A “road server” 25 milliseconds away can act like a game server, defining an environment for the car but not maintaining constant control of the brakes. 

Hoffman is remarkable, one of very few in this space who can “think different.” Nearly everyone else says, “This is our 5G network, better and faster than our 4G. These are user cases that fit.” Without innovative ideas like Hoffman’s, those “use cases” may never work well enough. 

Wired called Hoffman a pioneer in cloud computing. He’s an irrepressible character with a PhD in Molecular Pathology who once led a large part of Ericsson and has depth in the technologies. He believes technology should be, “Unique. Easy to use. Zero barrier of entry. Obvious when you see, not obvious when you haven’t.” 

What could be more important on the Internet than Pokémon?

Improvements welcome