The U.S. Federal Communications Commission (FCC) is requesting input from the public on new technological approaches to assessing “real-time, non-Federal (government) spectrum usage, so that it has better insights into current technologies that might help the agency to manage spectrum and identify opportunities for spectrum sharing—including how artificial intelligence (AI) might be used.
This FCC Notice of Inquiry (NOI) was approved by all four members of the Commission. It states:
“Spectrum usage information is generally non-public and made available infrequently. As the radiofrequency (RF) environment grows more congested, however, we anticipate a greater need to consider such data to improve spectrum management. That is especially true as the burgeoning growth of machine learning (ML) and artificial intelligence (AI) offer revolutionary insights into large and complex datasets. Leveraging today’s tools to understand tomorrow’s commercial spectrum usage can help identify new opportunities to facilitate more efficient spectrum use, including
new spectrum sharing techniques and approaches to enable co-existence among users and services.”
Spectrum usage has been defined in various ways. In one technical paper, for instance, NTIA and NIST defined “band occupancy” as “the percentage of frequencies or channels in the band with a detected signal level that exceeds a default or user-defined threshold.”
“Right now, so many of our commercial spectrum bands are growing crowded,” said FCC Chairwoman Jessica Rosenworcel. “Hundreds of millions of wireless connections—from smartphones to medical sensors—are using this invisible infrastructure. And that number is growing fast. But congestion can make it harder to make room in our skies for new technologies and new services. Yet we have to find a way, because no one wants innovation to grind to a halt. To do this we need smarter policies, like efforts that facilitate more efficient use of this scarce resource. … Now enter AI. A large wireless provider’s network can generate several million performance measurements every minute. Using those measurements, machine learning can provide insights that help better understand network usage, support greater spectrum efficiency, and improve resiliency by making it possible to heal networks on their own.”
“[This] inquiry is a way to understand this kind of potential and help ensure it develops here in the United States first. “I believe we can do more to increase our understanding of spectrum utilization and support the development of AI tools in wireless networks,” she added.
Rosenworcel noted that some pioneering work on dynamic, cognitive radios was kick-started with the Defense Advanced Research Project’s three-year Spectrum Collaboration Challenge, which sought to develop software-defined radios’ capability to dynamically detect other spectrum users and work around them in a congested radio frequency environment.
The FCC pointed out in a statement that it generally doesn’t collect information on spectrum usage, and instead relies on intermittent data from third-party sources.
“As the radiofrequency environment becomes more congested, leveraging technologies such as artificial intelligence to understand spectrum usage and draw insights from large and complex datasets can help facilitate more efficient spectrum use, including new spectrum sharing techniques and approaches to enable co-existence among users and services,” the agency said, adding that the inquiry will explore the “feasibility, benefits, and limitations” of various ways to understand non-federal spectrum usage, as well as band- or service-specific considerations and various technical, practical or legal aspects that should be considered.