Hacker News Reader

Comment by Analemma_ | Hacker News Reader

Phones and other devices with cellular radios still use differential GPS by way of the cell towers (which have known, fixed positions) to do those corrections. You can notice this because, when you don't have a cell signal, your phone is considerably slower to get a GPS lock and less accurate when it does. Presumably these farms are out in the middle of nowhere and so that's why they can't do the same thing and need a satellite-based augmentation instead.

View on HN

Not quite. Phones use assisted GPS, which means that they download GPS ephemeris data over the internet instead of getting it from the the satellite signal (because the satellites transmit data at 50 bps). So the time to first position is faster.

Differential GPS, by contrast, transmits the difference between the surveyed location of a base station and the GPS-derived position of the same station. That offset can be used locally to correct other nearby GPS receivers.

Phones also use trilateration on the cell tower signals to get an approximate solution before trying to solve 3d+time on the GPS signal.

It makes me unreasonably happy to see someone call it trilateration. Then again, you appear to know what you’re talking about.

They also use databases of know wifi hotspots as a way to bootstrap location when in difficult GPS environments like urban canyons.

And yet my uber driver still can't find my house.

Most modern cellphones use AGPS, assisted GPS, to reduce the time to get their first fix. Basically, AGPS primes the GPS receiver with an accurate time of day and the GPS almanac (satellite ephemeris data). That's rather different from Differential GPS, which is basically corrections used to compensate for fluctuating conditions in the atmosphere.

I may be wrong, but I don't know of any system for cell-tower derived corrections being redistributed out as GNSS corrections. Could you clarify the specific system you described?

Not so much. Modern GPS will get a first fix pretty quickly regardless, you don't need to seed it.

AGPS is used when satellite coverage is poor, like urban canyons in a city, also a situation where there are a lot of cell towers and triangulation is somewhat useful

AGPS refers to a specific set of technologies, MSB where the almanac and time seed is provided over the cellular network, and MSA where the unit sends GPS data to the AGPS server and the server solves for location using whatever data it would like (which can include a multitude of corrections).

I think what you're referring to with cell tower triangulation (and also WiFi location, etc.) is generally referred to as "hybrid location service" and is usually done at a higher level (ie - in software) using AGPS as one input to a sensor fusion algorithm.

The time to first fix is mainly impacted by whether the receiver knows which satellites to listen to. This requires two things: having the ephemeris for the constellation (either 10min of listening to GPS signal, or getting that through AGPS), and it's approximate own position/time (done through cell antenna locations, and NTP) to know which satellites should be visible.

Once this is available, the receiver can then focus on the signals that are expected and will then be able to provide a 3D fix as soon as it receives something from 4 satellites, which would be in a few seconds.

A-GPS can run in Mobile Station Assisted (MSA) mode, where the device sends raw GPS data to the A-GPS network and the assistance server processes it. In this situation, the assistance server can use whatever precise cell site location and ionospheric correction data it wants in its processing, although this is more of an "exercise left to the implementer" and I'm not sure sure MSA mode is even very common anymore.

I think was only a brief window where MSA was used, when phone processors were weak and cheaper to have limited GPS chip. Now, GPS chip do everything onboard and are included on radio processor or main processor. Plus, phone processors are powerful enough to do GPS themselves.

> Presumably these farms are out in the middle of nowhere and so that's why they can't do the same thing and need a satellite-based augmentation instead.

Easily maybe.

I can’t see any reason someone can’t take a stationary GPS enabled device and use that to broadcast the differential between received location and actual location. Probably could do it for less than $20 in equipment. Add multiple units and get triangulation for free.

Australia has a reasonable amount of yearly drift. It currently moves at about 6.8cm (2.7in) per year.

That is enough movement that it can throw off purely GPS solutions.

https://www.australiangeographic.com.au/topics/science-envir...

> ...

> But two important things have happened since then. Australia has moved about 1.6 metres northeast, effectively moving the location of mapped features and their associated GDA94 coordinates.

> At the same time, positioning technology has evolved considerably. By 2020, Australia will have moved by 1.8 metres and many of us will own devices that could pinpoint locations with accuracy as small as, well, a smartphone. With real-time access to precise satellite positioning at our fingertips, we’ll notice discrepancies with GDA94-mapped features.

> Imagine the confusion if a driverless car can determine its coordinates with (say) decimetre accuracy, in the GNSS datum, but the stored map features were referenced to our old datum GDA94 and and were nearly two metres different. That’s probably not even in the same lane.

As I understand it, RTK nulls that out. If your RTK base is fixed to the landmass that's moving, your rover positions will move right along with it. So a farmer with an RTK base stuck atop their silo or whatever should be fine.

Cell towers have no relation whatsoever to any of the different types of differential GNSS

Reply on news.ycombinator.com