Qualcomm Patents a GPS System That Filters Out Satellites With Blocked Signals
GPS often goes wrong not because the satellites are broken, but because their signals bounce off buildings before reaching your phone. Qualcomm's new patent targets exactly that problem by automatically identifying and discarding those unreliable signals before they skew your location.
How Qualcomm's GPS filter handles blocked satellite signals
Imagine you're navigating a city and your GPS puts you half a block away from where you actually are. That usually happens because the signal from a satellite overhead bounces off a skyscraper before it reaches your phone, arriving a tiny bit late. Your phone measures that delay and gets the math wrong.
Qualcomm's patent describes a system that actively watches for this. It counts how many satellites have their signals blocked or bounced (called "non-line-of-sight" satellites), then calculates how much location error those bad signals are likely adding. If the error from a particular satellite exceeds what the system already thinks your position uncertainty is, that satellite gets dropped from the calculation.
The result is a cleaned-up list of only the satellites whose signals arrived directly and cleanly. Your phone then uses only those signals to calculate where you are, which should produce a tighter, more reliable location estimate.
determine a number of non-line of sight (NLOS) satellite vehicles (SVs), of a plurality of SVs, transmitting a plurality of NLOS signals; generate an updated horizontal uncertainty for the device based on the number of the NLOS SVs; …
Translation: The system counts blocked satellite signals and recalculates how uncertain it is about the device location.
How the system scores and drops unreliable satellites
The patent describes a processor-level routine that runs during GPS positioning sessions. Here's the core sequence:
- Count blocked satellites: The system identifies which satellites in view are transmitting "non-line-of-sight" (NLOS) signals, meaning the signal bounced off a surface rather than traveling in a straight line from sky to device.
- Update the uncertainty estimate: Based on how many NLOS satellites are present, the system recalculates the device's expected horizontal position uncertainty, basically widening or narrowing its confidence window.
- Measure individual errors: For each NLOS satellite, it calculates that satellite's measurement error. If the error is larger than the device's previous uncertainty estimate, the satellite is flagged as unreliable.
- Rebuild the satellite list: Flagged satellites are removed, producing a cleaned "updated SV list." The final position is calculated using only signals from satellites on that list.
The patent also references session timing, noting that the system checks how much time has passed since the last positioning session. This suggests the filter can carry forward known-good satellite data from recent sessions rather than starting from scratch every time.
remove the one or more NLOS SVs from an SV list comprising the plurality of SVs to generate an updated SV list.
Translation: Satellites with blocked signals are dropped from the active tracking list to improve accuracy.
What better GPS filtering means for your phone's location
GPS accuracy in dense urban areas is a persistent problem, and most current approaches handle it after the fact by applying statistical corrections to an already-computed position. Qualcomm's approach is more surgical: it removes the bad inputs before the position calculation runs, which is a cleaner fix.
For your phone, this could mean fewer moments where navigation apps place you on the wrong side of a street or inside a building you're standing next to. Qualcomm supplies GPS and positioning chips to a huge share of Android devices worldwide, so improvements at this layer tend to show up broadly across manufacturers.
This is practical, unsexy engineering work on a problem that genuinely frustrates people every day. It's not a moonshot, but better urban GPS accuracy is something most smartphone users would notice immediately. The real question is how much improvement Qualcomm can deliver in practice, since NLOS filtering is a well-studied problem and the gains from any single technique tend to be incremental.
There are more where this came from
We read every patent application Big Tech publishes and send you the ones worth knowing. Plain English, free, every week.
The drawings
10 drawing sheets from US 2026/0227527 A1 · click any drawing to enlarge
Want this weekly breakdown for a company we don't cover? Patentlyze Pro →