New Initiative Aims to Ensure 5G Networks Are Reliable, Secure

McKeown’s group invented SDN to solve technical and commercial problems that had begun to crop up, first on wired networks like the early internet, but later on cellular and wireless as well, as those networks began to proliferate.

All this information must flow like water through pipes, but in this case, the pipes are physical wires or wireless channels. The goal in networking is simply to not fail – maintaining the flow of data even in the face of a nuclear attack. To do that, computer scientists developed a technology that parsed big buckets of information, such as text, images, music or streaming movies, into gazillions of smaller droplets of data known as “packets.”

The network essentially has two tasks: First, the data packets must be addressed and forwarded toward their intended destinations and reassembled into their original form. Second, the data must get routed through the network by way of wires or wireless channels – the pipes in this analogy. If one pipe is slowed or clogged, the router simply chooses a different pipe.

But, as data traffic exploded over the subsequent years, and more and more packets coursed through these channels, router manufacturers started adding proprietary software to their once-simple routing boxes. “You had barnacles upon barnacles of inscrutable code clogging up the routers, making it difficult for network operators to fix data interruptions when they occurred,” McKeown said.

In 2007, Martin Casado, then a Stanford graduate student and now a Silicon Valley venture capitalist, wrote a seminal paper proposing to create software-defined networks – virtual plumbing that scrapped the proprietary code for open source programs. Suddenly, network operators could control the data flow, remotely, all the way from Point A to Point B, and relegate the routers back to their original job of merely reading the addresses off packets and sending them on their way.

Internet companies, chip makers and other network stakeholders quickly got behind SDN, working together to create the necessary hardware and software – like the P4 network control software – that enable cloud computing operations to manage ever-growing data flows with rarely a blip. Today, this paradigm faces a new hurdle: the fact that the makers of these new 5G wireless networks are no longer headquartered in America, but in China and Europe.

“For the first time in history, there is not a single U.S. manufacturer of cellular telephone equipment. Meanwhile, the world is building 5G infrastructure on equipment where you have no idea what’s in the boxes,” McKeown said. “This is DARPA’s worry. This is the government’s worry. And they should be worried.”

Against that backdrop, roughly two years ago DARPA solicited the research proposals that coalesced into Project Pronto. The demo on the Stanford campus is a proof of concept that SDN systems, adapted to work on 5G networks, can thwart hacks on drones flown by the lab of Stanford aeronautics and astronautics professor Mac Schwager, all in under a second – far quicker than the minutes or hours it might take today.

This first test will be fairly simple: When the computer scientists flip on the SDN shield, the drones should fly true through the attack. When they toggle off the protection, the craft should crash to the ground or collide. “We will smash up a few drones, but fortunately they’re fairly robust,” said McKeown, who is to receive the IEEE Alexander Graham Bell Medal for his continuing contributions to network technology.

Project Pronto’s second and longer-term goal will be to demonstrate that experimental SDN systems at each of the collaborating universities can also run 5G network test beds. Here, the university researchers are working with dozens of industry players – cloud service companies, chip makers, data security vendors, and network traffic carriers – brought together through the ONF.

ONF will translate SDN research from the universities into wireless network management protocols that would have an important intellectual property characteristic – they would have no IP at all thanks to their open source development model. Like the original internet, open source adheres to simple rules. Anyone is free to download any open source product, and to modify and improve upon the product as they wish, so long as they throw any modifications or improvements they make back to the open source community for further adaptation or refinement.

“I think it’s this combination of the open source ethos and the deep programmability of SDN that will make future wireless networks more reliable and more secure,” McKeown said.