Researchers Turn to "Sidewalk Matching" to Fix the GPS Urban Canyon Problem, Improve Accuracy
A new approach to GNSS location calculations in urban areas delivers a major gain in accuracy — using existing smartphone technology.
Researchers from Shenzhen University, the Hong Kong Polytechnic University, and Wuhan University have come up with a way to boost the quality of GPS and other global navigation satellite system (GNSS) systems for pedestrians in urban environments — working around the "shadows" cast by skyscrapers by simply following the sidewalk.
"This sidewalk matching technique represents a significant leap forward in urban pedestrian navigation," claims lead researcher Duojie Weng of the project. "By combining smartphone sensors with pedestrian network data, we’ve created a practical solution that doesn't require costly 3D models or large training datasets. This makes it an accessible, cost-effective solution for a wide range of real-world applications, from pedestrian safety to ride-hailing services."
The issue the team set out to solve is known as the "urban canyon" problem: when GNSS users in urban environments walk near tall buildings like skyscrapers and lose line-of-sight to the satellites providing the timing information for their position. Combined with radio reflections from the same buildings, it's enough to confuse a GPS receiver — as anyone who's seen their location bounce around a few-hundred-feet circle downtime will attest.
The researchers' solution: using smartphone sensors and pedestrian mapping data to perform "sidewalk matching. In this, the receiver calculates that half of the sky has line-of-sight to more satellites, as a way of determining which side of the road the user is on, and then filters out faulty measurements using a sliding-window system based on the carrier-to-noise ratio and azimuth angle data.
Combined with data from the accelerometer and gyroscope in the smartphone to provide dead-reckoning movement information, the result is a dramatic improvement in accuracy: tests in Hong Kong's skyscraper-strewn streets accurately tracked the users' positions to below 16 feet, a big gain in accuracy over existing approaches.
The team is hoping to see the approach used beyond simple phone-based navigation systems, proposing for collision avoidance systems, ride-hailing services, and accessibility for the visually impaired.
The team's work has been published in the journal Satellite Navigation under open-access terms.
Freelance journalist, technical author, hacker, tinkerer, erstwhile sysadmin. For hire: freelance@halfacree.co.uk.