Your phone works out its location by timing radio signals from satellites and doing the math on how long each one took to arrive. That’s the whole trick behind GPS: distance from timing, timing from light speed, and location from distance.

How a phone figures out where it is

At least 24 GPS satellites (the U.S. Space Force currently flies 31) orbit roughly 20,200 kilometers up, each one circling Earth about twice a day. Every satellite carries onboard atomic clocks accurate to within nanoseconds and constantly broadcasts its own position and the exact time the signal was sent. A phone’s GPS chip never talks back to the satellites — it only listens, then compares the timestamp riding on each incoming signal to its own internal clock.

Because radio waves travel at the speed of light, about 299,792,458 meters per second, the tiny delay between when a signal left a satellite and when it arrived tells the receiver exactly how far away that satellite is. This is called trilateration: calculating a position from distances to several known points, the same way you could narrow down where you’re standing if three people at known locations each told you exactly how far away you are. One satellite narrows a phone’s location to a giant sphere. A second narrows it to a circle. A third brings it down to essentially one point on Earth’s surface. A fourth is what makes the whole system usable day to day, because it lets the receiver correct for its own clock error — ordinary phone clocks are nowhere near as precise as the atomic clocks overhead, and that fourth measurement cancels the difference out.

Why the satellites need Einstein’s physics

None of this works unless the system also corrects for relativity. Satellites moving at roughly 14,000 kilometers per hour experience special-relativistic time dilation, which makes their clocks run slightly slower than clocks on the ground. At the same time, they sit in weaker gravity far from Earth’s mass, and general relativity says that makes clocks run faster. The gravitational effect wins out overall, so engineers deliberately tune GPS satellite clocks to tick slightly slow before launch, so that once in orbit they end up matching time on the ground. Skip that correction and position errors would pile up by about 10 kilometers a day — enough to make the system worthless within hours.

What actually limits GPS accuracy today

Consumer GPS chips are typically accurate to within a few meters outdoors, though the raw signal loses some precision on the way down. Radio waves slow slightly as they pass through the ionosphere and troposphere, and they can bounce off buildings before reaching a receiver — a problem called multipath, where a signal arrives twice by two different paths and confuses the calculation, which is why GPS gets unreliable in dense cities or narrow canyons. Correction techniques such as differential corrections and dual-frequency receivers, which compare two GPS frequencies to estimate and cancel out atmospheric delay, have pushed everyday accuracy well past what the original 1970s design ever promised.

None of this would matter much if a few meters of error didn’t matter in practice — but it’s the difference between a map telling you which street you’re on and which lane, between a delivery drone clearing a rooftop and clipping it, between an aircraft’s autopilot trusting its position and not. That’s why accuracy, not just coverage, is what every GPS upgrade has actually been chasing.

GPS isn’t the only satellite navigation system in use. Russia’s GLONASS, the European Union’s Galileo, and China’s BeiDou all offer comparable global coverage, and most modern phones quietly draw on signals from several of these constellations at once, locking onto a position faster and more precisely than any single system could manage alone.