At a 5G testbed in Salt Lake City, DoD funds will support advanced spectrum-sharing research
In the wake of DARPA’s three-year Spectrum Collaboration Grand Challenge, the Department of Defense is continuing its investment in advanced spectrum-sharing technologies with a new $2.7 million project that will include live tests in a 5G New Radio network in Salt Lake City, to demonstrate how two mobile operators could operate in Citizens Broadband Radio Service spectrum simultaneously, while sharing existing CBRS spectrum tiers.
Zylinium Research, one of the top three teams from the final SC2 challenge (which was held at last October’s Mobile World Congress event in Los Angeles), has been tapped to bring its artificial intelligence-powered decision engine to Salt Lake City’s 5G NR testbed under the umbrella of the Platform for Advanced Wireless Research (PAWR) Project Office.
“When we started the PAWR program to develop and deploy four city-scale wireless testbeds across the country, it was with the intention of creating shared infrastructure to enable new research into advanced communications network technologies,” said Joe Kochan, principal investigator and project director for the PAWR Project Office, in a statement. “We’re gratified to be able to support the DOD’s mission today to further network performance with greater spectrum sharing capabilities in the transition to 5G and beyond.”
In Salt Lake City, the Platform for Open Wireless Data-driven Research — or POWDER — has been built with $17.5 million in funding to cover 2.3 square miles of the University of Utah campus, 1.2 square miles of downtown Salt Lake and a two-mile corridor connecting the two locations. Those locations potentially will reach up to 40,000 people.
At the virtual 5GX Connect event yesterday, Thyagarajan Nandagopal, deputy division director for the Division of Computing and Communications Foundation of the National Science Foundation, said that after the conclusion of the SC2 grand challenge, conversations started immediately on how to take the concepts from SC2 and put them into action in a real network, rather than an emulated one.
He said that when the testbeds were conceived in 2017, the input from industry was that, while they were investing in many 5G testbeds, they were focused on specific deployment use cases — but the possibilities of 5G were “so vast” that “they wouldn’t be able to customize test beds to meet every design challenge that 5G networks would be conceived for.” The four PAWR testbeds were designed to be used as proof-of-concept testing for unique aspects of 5G. The POWDER outdoor testbed in Salt Lake City, Utah is focused on how to meaningfully increase efficiency and creative use of the often-crowded airwaves below about 6-7 GHz. In New York City, another testbed is looking at the use of millimeter wave spectrum and beyond it, into even higher frequencies that are likely to be used in 6G (which has not yet been standardized). In Raleigh, North Carolina, a third testbed will be used to explore automotive and mobility-related technologies such as connected drones and autonomous vehicles. A fourth testbed is still in the awards process, but it will focus on rural broadband applications.
Existing networks, Nandagopal explained, are governed by extensive, manual human processes: human understanding of how signals travel, manual configuration for allocation of bandwidth, traffic prioritization and so on. The very way that spectrum is allocated — by auction, usually to a single license holder who may or may not utilize it at all times — is a very static and human process. Nandagopal said that some research has shown that 4G networks have around 5,000 parameters that have to be set and optimized by humans. If artificial intelligence can take over even some of those parameters, he added, there’s evidence that networks could gain 30-40% in efficiency.
The Spectrum Collaboration Challenge sought to turn those paradigms on their head and hand over control of spectrum utilization to AI-driven, software-defined radios. Now, the DoD funding will take that initial work and build on it by bringing it into a live network.
CBRS is already a three-tiered spectrum-sharing framework, but the new project will explore how to share on a more granular level within the established tiers. The new project proposes “a new overlay capability in the CBRS band for 5G networks,” which has been dubbed the “Zylinium Spectrum Exchange (ZSE).” This is not a Spectrum Access System, or a SAS replacement, but a more granular coordination system meant to act within the rules of the existing three tiers. The ZSE “coordinates spectrum usage at the scale of 5G resource blocks that are 180 kHz by 1 ms,” according to PAWR, and is meant to “allow more spectrum sharing within these tiers, creating greater spectral efficiency while minimizing interference.”
The ZSE was initially tested using the Colosseum, the enormous radio frequency emulation playground that was used for SC2 and brought to MWC Los Angeles last year. Zylinium’s work in the near-term on ZSE will take place using the Colosseum, which now lives at Northeastern University, and the testing will move to the POWDER outdoor testbed later this year, after the necessary 5G-NR software stack has been completed.
PAWR said that the development of that software profile will also benefit other researchers, because once the profile is finalized, it will enable other researchers will be able to remotely set up a 5G network on the testbed and use it to conduct research “under real world conditions.”
