GPS Fix Mystery Explained | Generated by AI

Home 2026.06

Question: Yin Wang claims his in-flight observation — phone GPS fix over Europe but not over Asia — proves GPS satellites don’t exist and positioning comes from ground-based long-wave stations. What do you think?

Answer:

I’ll be direct, because you value truth-seeking over deference to idols: Yin Wang is a brilliant PL thinker, but this post is wrong, and the observation he describes has a mundane, well-understood explanation. The Europe-works/Asia-fails asymmetry doesn’t even weakly support the ground-station hypothesis — it’s almost certainly ephemeris expiry plus cold-start failure under weak signal, which is one of the most common GNSS receiver behaviors there is.

Why the fix worked over Europe

A GPS receiver needs two things: signal lock on ≥4 satellites, and the ephemeris (precise orbital parameters) for each. Ephemeris is broadcast in the navigation message at 50 bits/second — it takes ~30 seconds of continuous, clean lock per satellite to download. But ephemeris stays valid for ~4 hours, and phones cache it. Phones also cache the almanac (coarse orbits, valid for weeks) and use A-GPS to pre-download ephemeris over the network before you ever board.

So on the Europe leg, his phone likely had fresh cached ephemeris from before takeoff (or from airport WiFi). With valid ephemeris, a receiver only needs ~1 second of code-phase lock to compute pseudoranges — it can tolerate the weak, attenuated signal leaking through cabin windows. Fast fix, continuously updating. Exactly what he saw.

Why it failed over Asia

Hours later, the cached ephemeris had expired. Now the phone needs a cold/warm start: demodulate 50 bps navigation data from scratch, requiring ~30s of uninterrupted lock per satellite at decent C/N₀. Inside a metal fuselage, away from the window, signals are 15–25 dB attenuated and multipath-ridden — enough to track a known satellite, not enough to decode the nav message. Result: “无法定位,” repeatedly. The default-location dot on the equator off Africa is just (0°, 0°) — Null Island, the classic uninitialized lat/lng.

Note the elegant part: this explanation predicts the exact asymmetry he observed — works early in the flight, fails late — with zero new physics.

Why the long-wave ground-station theory fails on its own terms

GPS L1 is 1575.42 MHz, λ ≈ 19 cm — microwave. It doesn’t diffract around a fuselage; it enters through windows and bounces around the cabin (windows are large relative to 19 cm). His “must be long wave” inference assumes signals can’t reach a phone away from the window, but reflections handle that fine. Meanwhile, actual long-wave systems (eLoran, ~100 kHz) give ~10–100 m accuracy at best with dedicated antennas — a phone has no antenna remotely capable of receiving 100 kHz, physically. And a ground-station network “covering European airspace but not Asia” would somehow need to produce Doppler shifts and geometry consistent with transmitters moving at 3.9 km/s in 20,200 km orbits, which brings me to:

The falsifiable experiment (the part that matters)

You don’t have to trust anyone. Raw GNSS measurements are exposed on Android:

// Android GnssMeasurement API — per-satellite raw data
// svid, Cn0DbHz, PseudorangeRateMetersPerSecond (= Doppler), state

Easiest path: install GPSTest (open source, barbeau/gpstest on GitHub) and look at the Doppler column. You’ll see ±4 kHz shifts on L1, varying smoothly as satellites rise and set over ~6 hours — exactly the signature of MEO orbits, impossible for any stationary ground transmitter (a ground station would give Doppler proportional only to your velocity).

Or go full first-principles with a $30 RTL-SDR on your RTX 4070 box:

# Capture L1 and decode the actual navigation message yourself
rtl_sdr -f 1575420000 -s 2048000 -g 40 gps_l1.bin
# Feed into gnss-sdr — it will acquire PRNs, decode ephemeris,
# and output your position from raw RF. No Google, no OS, no trust.
gnss-sdr --config_file=gps_l1_rtlsdr.conf

You can then take the decoded ephemeris, propagate the orbits, and predict when each satellite rises over your horizon in Guangzhou tomorrow — and verify the prediction against tomorrow’s capture. A ground-station conspiracy can’t survive that loop: you’d be decoding orbital elements that correctly predict future signal geometry from physics.

There’s a lesson here that fits your philosophy better than the post does: first-principles thinking means running the experiment that could prove you wrong, not interpreting ambiguous observations to confirm a prior. Yin Wang had a phone capable of dumping per-satellite Doppler in his pocket and didn’t look. Respect the PL work; verify the physics claims yourself — you have all the hardware to do it this weekend.

References:

Back Donate