The effort to rebuild the internet so it can handle the anticipated onslaught of connected devices and people has reached a new phase in the U.S., now that the National Science Foundation is giving three research institutions a total of $15 million to take their ideas from paper to pilot.
The goal here is to build out a physical networking infrastructure that can handle more devices in a flexible, secure way without introducing latency, vulnerabilities and giving too much control to any one participant. The three research projects cover vehicle networks that we might need for driverless cars to become reality, real-time building management applications, medtech and new ways of delivering content. But all offer a new method for addressing on the internet — switching from IP addresses to naming the devices based on their context or content.
Let’s take a deeper look.
XIA : A select number of cars on the streets of Pittsburgh and an undisclosed video delivery network will get to try out the eXpressive Internet Architecture (XIA). This project, led by researchers at Carnegie Mellon University, with partners at the University of Wisconsin-Madison, Duke University and Boston University, focuses on security and trustworthiness between devices communicating on the network. The pilot projects are designed to continue testing the core elements as well as add mobility and a better understanding of how to develop the mechanism to open and close the sessions created between devices during packet delivery.
Named Data Networking: The idea behind Named Data Networking is that if we assume the internet is primarily about delivering information to people, then instead of relying on esoteric IP addresses (a focus on where) we should use names for the devices (a focus on what). This helps route data more effectively and flattens out the current model of the internet when a connected light switch in your home might have to route a request back to the cloud just to turn on a connected bulb.
The researchers at the University of California, Los Angeles, along with several other university research teams, will use their grant to deploy NDN systems at Open mHealth, a non-profit patient-centric ecosystem and at UCLA Facilities Management which operates the second largest Siemens building monitoring system on the West Coast. The idea will be on figuring out how to design applications to take advantage of this network architecture, understand security, how to route and forward data and figure out how to build storage that fits this model.
The MobilityFirst Project: What happens when you apply a named data network in a mobile situation? That’s the premise behind the trials of the MobilityFirst architecture, which will involve a wireless ISP called 5Nines in Madison, Wisconsin; the creation of a content production and delivery network involving several public broadcasting stations in Pennsylvania connected by a greenfield optical network called PennREN; and a context-aware public service weather emergency notification system with end-users in the Dallas/Fort Worth area.
The trials are led by Rutgers University with partners at the University of Wisconsin-Madison, Massachusetts Institute of Technology, University of Michigan, University of Nebraska-Lincoln, University of Massachusetts-Amherst and Duke University. The key element of this network appears to be a storage-aware routing technique that the proposal said provides “significant improvements in mobile network capacity and functionality.”
Many of these trials are trying to figure out how these new network architectures can work alongside existing content and applications, as well as understand how the new networking stacks will relate to the existing OSI networking stack that governs how everything works in IP today. There’s a lot of emphasis on programmability, be it via software defined networking or related to storing data at various nodes in a network based on any number of factors from security to who is demanding the content.
Perhaps what is most striking here is how software is taking over more and more of the physical functions of hardware as our hardware must scale out and become more complex. Obviously, hardware will determine the underlying performance one can use in setting software-defined networks or storage, but as the cloud has proven, you can offer more security and flexibility if you abstract some of the cabling and ports in software. As the internet grows, such traits will increasingly become important.