How Cosmic Rays and GPUs Can Help Detect Concealed Nukes

by Tony Kontzer

Building from groundbreaking work by Los Alamos National Laboratory, researchers at Decision Sciences are tapping the invisible cosmic rays passing through us to detect threats that we can’t see.

Thousands of particles called muons constantly pass through us, and a team led by Michael Sossong, chief physicist at Decision Sciences, is working on a commercially viable way to measure their trajectories as they pass through nuclear materials, weapons of mass destructions, or  illegal contraband.

Doing so, Sossong told a room of attendees at NVIDIA’s GPU Technology Conference, will help security officials conduct more revealing and accurate scans of airline and maritime cargo, as well as vehicles crossing land borders.

“We want to come up with technology to be able to see inside all of these containers to know if someone is trying to smuggle in nuclear warheads,” Sosson said.

Decision Sciences has invested millions since 2006 on developing a commercial solution for so-called “muon scattering tomography,” which measures the way muons’ paths change when they encounter certain materials.

Muons are 200 times heavier than electrons, and thus aren’t impacted by collisions with them. However, as muons interact with denser particles, their paths change — a phenomenon known as “scattering.” By reading measurements of this scattering, scanners can detect pretty much any substance.

Decision Sciences has been testing the technology with a vehicle scanner at a port in the Bahamas, scanning both actual vehicles and dummy test vehicles, although Sossong wouldn’t reveal the substances the test scans have detected. He also didn’t offer a time table for the introduction of a commercial product.

Where GPUs come into play is making the scans faster – a critical consideration if the technology is to find success in the marketplace. Sossong said that images have to be reconstructed in real time, enabling scans every five seconds to keep up with the flow of trucks passing through a border.

While CPUs take about 40 seconds to process the image data, GPUs have accelerated performance by 100-fold in analyzing the split-second collection of data on 10,000 muons passing through objects every second.

Sossong said that other applications for the technology include using muons to monitor the atmosphere to detect weather patterns; perform overnight low-radiation CT scans of cancer patients who can’t be subjected to the concentrated radiation of traditional CT scans; and scan bridges to detect developing  structural weaknesses.