The Smith Institute has won a Queen’s Award for International Trade! It was tricky to keep the news to ourselves before the embargo lifted, but now we can finally put this prestigious award on display and talk to you about how it came to be in our hands.
For outstanding achievement in international trade, the accolade was awarded to the Smith Institute in recognition of our world-leading work in spectrum auction verification.
Radio spectrum is used by any device that communicates wirelessly. This includes televisions, radar, satellites, baby monitors, radio broadcasts, remote controls and mobile phones. Different bands of spectrum are allocated for different purposes. These bands need to be managed carefully to avoid interference, ensuring people, businesses and networks are not disrupted.
The part we played (the short version)
Beginning with the UK, the Smith Institute’s work grew to involve regulators in 10 different countries. We were brought in to check that spectrum bands were distributed to different mobile networks via auction in a verifiably fair manner that encouraged market competition.
This wasn’t a small challenge. To date, the Smith Institute has quality-assured the algorithms underpinning the auction software of 30 highly complex auctions, which together have raised many billions of pounds. Recently a landmark FCC auction for 5G mid-band spectrum in the USA raised $80bn. It made more spectrum licenses available in a single FCC auction than ever before.
But our story didn’t begin with such grand figures or scale. It started because, with the 2008 auction in the 10-40GHz band, the UK decided to boost the use of auctions in finding a fair market price for spectrum – a step that would inspire other countries and their regulators to follow suit.
The first auctions and fair market price
Radio spectrum has been recognised as a valuable resource since the 1990s. Before that, it was often allocated without charge to mobile networks, which were in their early iterations at the time.
As demands have increased over the years – with the introduction of smartphones, 4G and now 5G – more and more spectrum has been made available to mobile networks. Regulators have been pivotal in enabling processes, such as auctions, that establish fair prices.
At the start of the new millennium, one of the first auctions in the UK was held for 3G spectrum. And to the surprise of everyone involved, the auction raised £22bn, far more than anyone had expected. There was a considerable amount of debate around whether mobile companies had overpaid.
The complexity of a combinatorial auction
The market price discussion was a complicated knot to untangle. Spectrum auctions are no traditional hand-raising and gavel-striking affair. They often sell many packages or ‘lots’ of spectrum licenses, none of which are independent of the others.
Simply explained, this is a little like trying to design an auction in which a bidder wants to win ornate chairs, but only if they can get the matching table… which is also paired with a set of matching table ornaments that someone else wishes to buy… but only if they can get the table too.
Both bidders value their preferred set of items more than they would pay for each individually, and they might not want to risk bidding on the table unless the chairs or ornaments are guaranteed.
Similarly, a mobile network’s business plan might be dependent on a certain number of spectrum licenses, also requiring certain parts of the bands on auction. Winning just one or two may therefore not be any good to the company. So who has won and what price they pay are two separate questions, and it requires a complex algorithm to work them out.
The test run that did what it said
Ofcom in the UK were one of the first regulators to realise this kind of combinatorial auction needed a specific method of implementation. So they used a small auction for non-mobile spectrum to test-run software designed to declare winners and prices fairly. But they needed a way of verifying that the software was working as they said it was.
They came to the Smith Institute and asked us to design scenarios to test and verify their spectrum auction. We worked out how to provide a certificate for Ofcom’s auction, having checked that their algorithm did what it said it would do. This provided them with the independent quality assurance they needed to roll out the auction software on a larger scale.
Other countries took note. And the Smith Institute were soon testing the algorithms underpinning spectrum auctions for the Swiss, then the Irish, and then other international regulators.
The complex journey abroad
Originally, when it came to running the verification tests, we would do it by hand, analysing the results manually. But once the Smith Institute went international, the scale and complexity of the auctions became far too great for this approach.
In many countries, there aren’t only spectrum licenses for different bands, but also for different geographical regions. The USA, for instance, regularly uses over 400 geographical regions for spectrum. This complicates the auction exponentially.
In such a scenario, to verify that a regulator’s algorithms were finding winners and prices as they were meant to, we had to invent ways of automating the tests we’d designed. Regulators now have many thousands of licenses up for sale in a single auction. And every time their software changes, even incrementally, we can rerun all tests automatically.
The cost of being incorrect is too large in a highly regulated industry like telecoms. That’s why our independent quality assurance is valued so highly by regulators across the world.
Since our first tests in the UK in 2008, we have become global leaders in this area. It’s been a fascinating journey and after years of hard work and innovation, it’s all the more rewarding to be presented with a Queen’s Award for International Trade.
Are you looking for independent verification of your systems and algorithms? We can provide you with the independent quality assurance you need to get peace of mind. Get in touch to talk to us about your system.
