Why GPS Speed Wanders in Cities and Dies in Tunnels

A GPS receiver works by timing signals from satellites and turning those times into distances. Everything that degrades accuracy does so by corrupting one of those distances — and in a city, the corruption has a specific and interesting cause.

Urban canyon

Between tall buildings, most of the sky is hidden. You might be receiving four or five satellites where an open road would give you fifteen, and the ones you can see are clustered in a strip overhead rather than spread across the sky.

This matters more than the raw count. When satellites are spread out, timing errors partly cancel. When they are clustered, the geometry amplifies them instead. Receivers report this as dilution of precision, and a high value means small errors become large position errors.

Multipath — the real culprit

Satellite signals bounce. A glass tower is an excellent radio mirror.

If the direct path to a satellite is blocked but a reflection reaches you, the receiver measures the reflected path — which is longer. It has no straightforward way to know this happened, so it treats that inflated distance as truth and computes a position that is wrong by the difference.

Worse, the reflection geometry changes constantly as you move. The error is not a steady offset you could correct; it shifts from second to second. And because speed is derived from how position changes, an error that moves looks exactly like motion. That is why the number wanders while you are sitting still at a city junction.

Tunnels and car parks

This case is simpler: the signal is gone. Concrete and earth stop it completely. The receiver has nothing to work with, so the reading freezes at its last value or drops out entirely.

When you emerge, reacquisition usually takes a few seconds, because the receiver still holds recent orbital data and knows roughly where it is. A recorded trip through a tunnel therefore shows a straight line from the entrance to the exit — no positions were produced in between. The distance is not badly wrong, because the straight line is close to the tunnel's actual path, but the route drawing is obviously simplified.

What actually helps

• Mount for sky view. The windscreen sees far more sky than the centre console, and much more than a cup holder. This is the single biggest lever you have.
• Wait a moment after stopping. The first seconds after a stop, and the first seconds after emerging from cover, are the least reliable. Do not judge the reading then.
• Prefer the map view to check plausibility. If the map shows you drifting off the road, you are looking at multipath, and the speed figure deserves the same scepticism.
• Trust averages over instants. A one-second error moves an average computed over twenty minutes almost not at all. This is why recorded trip figures are more dependable than any single live reading.

Setting expectations

Under open sky, on a motorway, a modern iPhone reports speed within a few km/h of truth, consistently. In a dense city core, the instantaneous figure can wander by more than that, and no app — including this one — can fix a signal that arrived by the wrong path.

What the app can do is show you what the receiver reports honestly, record trips in a form where transient errors get averaged away, and stay out of the way when the signal comes back.