How the FCC Achieved Excellence in Advanced Analytics and O.R.

Tuesday, May 1, 2018

For the past 46 years, INFORMS has hosted a competition to determine the Franz Edelman Award, whose purpose is to “bring forward, recognize and reward outstanding examples of operations research, management science and advanced analytics in practice in the world.” Considered the “Super Bowl” of the field, the competition for the award has a rigorous process for selecting the six finalists who make their final presentations to the judges at the annual INFORMS Business Analytics Conference. This year, the Federal Communications Commission (FCC) was the winner, and I had the pleasure and honor to be one of their coaches for their entry.

The judging committee looks at the following attributes of any project when they compare the different entries. All the finalists this year were strong, but I think the FCC did the best job of meeting these criteria.

  • Implementation. Is the work implemented and in use?
    • The FCC used advanced analytics to conduct a two-sided auction that concluded in 2017 to repurpose 84Mhz of TV spectrum for use by wireless carriers. The analytics were the key enabler that allowed this auction to be a success.
  • Impact/Value. What are the major quantified (e.g., dollars saved, revenue increased) and non-quantified (e.g., process streamlined, customer satisfaction improved) impacts of the work? How important was the work to the client?
    • The auction generated $19.8 billion in gross revenue from the wireless carriers, of which $10.05 billion was paid to broadcasters who were willing to give up their licenses. Another $1.75 billion was allocated to help move TV stations staying on the air to new broadcast frequencies. The net was a $7.3 billion reduction in the federal deficit.
    • Analytics were used to prove the viability of the auction to Congress, which led to the Spectrum Act of 2012. Congress passed this law with bi-partisan support that authorized the FCC to conduct the auction. I felt that this concept – having analytics studies influence the passage of a congressional law – was quite impressive.
    • By some estimates, the deployment of this new wireless capacity will increase productivity and stimulate the U.S. economy by many billions of dollars and hundreds of thousands of jobs.
    • Because of the success of the auction, the FCC has created a new office of Economics and Analytics, with new Chief Data Officer and Chief Data Scientist positions, that will be the center for the integration of Operations Research and Analytics into all aspects of decision making at the FCC.
  • Technical Solution. Is there a technical innovation in the project? Innovation may stem from the creation of new methods as well as the application of existing methods to new problems or new environments.
    • The FCC leveraged a number of different advanced analytics techniques. One major innovation was the use of heuristics, clique reformulation techniques, Bender’s decomposition, and distributed parallel solving integrated with a commercial mixed integer programming solver (Gurobi) to determine how stations could be reassigned to new channels while creating minimal new interference. Another was the use of machine learning techniques to determine the best combinations of a suite of satisfiability solvers to determine if reassignments were feasible during a part of the auction. To understand this particular business issue, the FCC determined for each 2-kilometer by 2-kilometer cell over the United States, the amount of interference that would exist between neighboring cells. Terabytes of data had to be processed to figure out the appropriate constraints that were used by the different solvers.
  • Difficulty. What political, technical, and managerial challenges had to be overcome in completing the project?
    • Both the wireless industry and the broadcasting industry were skeptical that the FCC could execute this auction. All processes, including documentation of the different mathematical models used, was published and made available for public comment. The industry provided feedback that shaped various decisions regarding the design and implementation of the auction.
  • Transportability. Is this work portable to other applications or industries?
    • Foreign governments are asking the FCC about how to use a similar auction that can help meet the demands for wireless coverage.
    • The same style of auction is being considered for reallocating other parts of the radio frequency spectrum for wireless use.

As a coach, I helped the team focus on telling their story about the problem they faced, how they approached it, and how their project met these different criteria. At Princeton Consultants, we try to help our clients understand all these issues to demonstrate the value of applying advanced analytics to their organization. Typically, the technical part is the easy part of the project. We help our clients handle the issues in the other categories, so they receive the maximum benefit from advanced analytics. Contact us to see how we can help you.