The math that keeps your phone connected two rooms away from the router was built to find something that turned out not to be there. In the 1970s, a group of radio astronomers went hunting for the radio signature of exploding black holes the size of a proton. They came up empty. What they did produce along the way was a signal-processing technique for pulling a clean signal out of a smeared, noisy mess, and that technique eventually became a patented piece of the Wi-Fi standard your router runs today. I have four access points in my ceilings and 1,200Mbps coming into the house. The chain from that failed hunt to the hardware overhead is shorter than you'd expect, and it explains a lot about how home networks behave.
A 1974 prediction sent radio astronomers hunting
Hawking said tiny black holes should leave a radio fingerprint
Stephen Hawking published work in 1974 arguing that black holes are not permanently sealed. Very small ones, he calculated, would slowly evaporate and finish their lives in a burst of energy that included radio waves. A black hole the size of a proton carrying the mass of a mountain would effectively announce its own death across the radio spectrum.
John O'Sullivan, an Australian engineer working on radio telescopes in the Netherlands, decided to go looking for one. Sensitivity was not the obstacle. Clarity was. A radio pulse crossing thousands of light-years of gas and dust does not arrive as a sharp spike. Different frequencies inside the same pulse travel at slightly different speeds, so the signal stretches into a low, wide bump and sinks below the background noise of the universe.
O'Sullivan and his collaborators built mathematical tools around fast Fourier transforms to reverse that stretching and rebuild the original spike. The tools worked. The black holes never turned up.
The math outlived the mission
Unsmearing a signal turned out to be a general-purpose problem
A fast Fourier transform takes a messy waveform and separates it into the individual frequencies that combine to make it. Once you know how each of those frequencies got delayed or distorted on the way in, you can work backward and reconstruct what was originally sent. Astronomers used it to sharpen telescope data. Nothing about it is specific to space.
By the late 1980s, O'Sullivan had returned to Australia and was heading signal processing at CSIRO's Division of Radiophysics. The agency wanted commercial applications for what its radio astronomy group already knew how to do, and wireless computer networking was an obvious target. Early attempts at it fell apart indoors. Speeds were poor, connections dropped, and nobody had a clean fix.
O'Sullivan recognized the shape of the problem. A signal ricocheting around a building gets mangled for a different reason than one crossing interstellar dust, but the receiver ends up staring at the same kind of mess. He pulled the astronomy math off the shelf. A CSIRO team of Graham Daniels, John Deane, Diethelm Ostry, and Terry Percival turned it into something a computer network could actually run.
Your living room is a small, noisy version of deep space
Reflections off drywall and appliances arrive at different times
When a US court examined CSIRO's patent in 2008, it laid out the indoor problem in plain language. Radio waves bounce off walls, furniture, and appliances, so one transmission reaches the receiver along several different paths and lands at several different moments. Engineers call this multipath. At slow data rates it barely registers. Push the rate up and the echo of one bit arrives on top of the next bit, and the receiver loses track of which is which.
US Patent 5,487,069 stacked three techniques to beat that. Data gets split across many parallel sub-channels, each running slowly enough that echoes fall harmlessly between symbols. Forward error correction repairs whatever damage still gets through. Bit interleaving scatters the errors instead of letting them pile up in one spot. The patent was granted on January 23, 1996, and it became part of both 802.11a and 802.11g.
Learning that changed how I think about my own network. Signal strength is the number everyone fixates on, but reflections are what quietly degrade a connection, which is why assigning each access point its own channel improved my speeds more than any hardware purchase, and why hardwired access points outperform a wireless mesh in a house full of interior walls.
Australia didn't invent Wi-Fi, and the real story is better
The patent covered the hard part, not the whole idea
The popular telling of this story ends with Australia inventing Wi-Fi, which is not what happened. CSIRO's own trial attorney told a jury in 2009 that the agency did not invent the concept of a wireless LAN, only the best method of making one work. Wireless networking was already under development in several places. Vic Hayes chaired the IEEE committee that hammered competing approaches into the 802.11 standard, and he gets called the father of Wi-Fi with roughly equal justification.
What CSIRO held was the piece that made indoor wireless fast and reliable enough to bother with, and it spent years collecting on that. Fourteen companies, including Intel, Dell, Microsoft, and Nintendo, settled in 2009 for a reported $205 million. A second group that included AT&T, T-Mobile, and Lenovo settled in 2012, adding roughly $229 million. Total royalties and settlements pushed past $430 million.
Solving the worst problem in indoor wireless is a narrower claim than inventing Wi-Fi. It also happens to be true, and it explains how the technique ended up inside nearly every access point sold since.
What a failed black hole hunt left in your ceiling
Research does not always deliver what it set out to find. O'Sullivan spent years chasing a signal nobody has detected to this day, and the tool he built for that hunt now runs in the firmware of every connected device in my house. My access points execute it constantly, sorting real data from its own echoes many times per second. That is a decent argument for funding curiosity with no obvious payoff. It is also a reminder that the physics fighting your network is the same physics that made it possible, which is why I still run Ethernet to anything that stays put.