Why Google Maps is WRONG (sometimes) – GPS Explained
You know when you are using Google Maps to navigate somewhere, and randomly it will show you driving on a completely different, often parallel road? Why does it do that? It’s using GPS to know where you are right, so why is it wrong? Well, let me explain a bit more about GPS, then talk about the tricks Google uses to get around the limitations, and ways to improve your GPS experience.
So, what is GPS? Well, that stands for Global Positioning System, and is the term most commonly used when talking about the American system, NAVSTAR. The American military created NAVSTAR in the late 1970’s, with the first satellite launching in February 1978. It was designed for a wide range of uses, but the key one was to allow intercontinental ballistic missiles, or ICBMs, to hit their targets accurately. The system was designed to have 24 satellites orbiting the earth once every 12 hours or so.
The USA started letting civilians use the platform in the 1980’s, but they weren’t alone up there. Russia, specifically the USSR launched GLONASS starting in 1982. They also have 24 satellites orbiting the earth, although thanks to them being about 1,000 km lower they orbit in 11 hs and 16 mins. Since then China, the EU, India and Japan have launched their own systems, although both the Indian and Japanese systems are in geostationary orbit for better accuracy in their own regions. The European system, called Galileo, has been operational since 2016 with more launches scheduled – although all 24 primary satellites are already in orbit and operational.
So, it’s a whole load of satellites. Does that mean your phone is communicating back and forth with them as they fly overhead? Well, not really. Your phone’s GPS module doesn’t transmit anything, there is no way you could fit a powerful enough transmitter into a phone to actively communicate with… well.. Anything in orbit, instead all it does is listen.
Each satellite is outfitted with an atomic clock, accurate to 1 nanosecond. Yeah. Nano. It broadcasts the precise current timestamp as it’s message, then your phone receives that message and can compare the current time to the timestamp embedded in the message and from that work out how far the satellite is. If you repeat that with two other satellites you can do what’s called triangulation – or in this case trilateration.
Two other key pieces of data that your phone has are called the Almanac and Ephemeris. The latter is stored in the message each satellite broadcasts and tells you where that specific one is on its orbit when it sent that message. The Almanac is general location information about all of the satellites – roughly where you should find them.
So, if you take the point where the first satellite said it was, then calculate the distance from it, that gives you a radius of where on earth you could be. Add a second satellite signal and you get a different radius that will overlap the first one. Add the third and wherever all three edges meet, that’s where you are. Now since you don’t have an atomic clock built into your phone, you actually need a fourth satellite to sync your clock and compare to. The more signals you receive and can compare the better a fix you will get.
Now, the GPS chip in your phone is doing this check once per second, or 1Hz. For the sorts of generally slow moving tracking the average person needs, this is decent enough, but it’s far from perfect. For example, if you are driving on a national speed limit road here in the UK like a motorway, you will have moved over 30 meters before your phone will check it’s location again. 30 meters is enough to miss a junction, and the faster you go the less accurate it will be.
On top of that, there is inherent inaccuracy in the GPS system. It’s only accurate to within 5 meters as standard, although enhanced units can be accurate up to 30 cm. GLONASS is accurate to between 2 m and 4 m, although Galileo is accurate to within 1 m for public use, or an insane 1 cm on it’s encrypted channel. Either way, you are likely to be between 1 m and 5 m off, hence why Maps regularly shows a circle around your location until it gets a confirmed fix.
So, if Google Maps is only getting location updates once per second, and those updates can be up to 5 meters off, how does it work as well as it does? The short answer: data. Google is a data company, they have a whole lot of it, and they leverage it to make Maps better. That starts with data from your phone’s accelerometer, if your phone doesn’t feel a significant change in G-Force, Maps can assume you haven’t changed direction or speed and it will smoothly update the animation of you continuing on the road it knows you are on.
To know which exact road you are on, of course, they use machine learning. They know what the speed limit of the road you are on is, they even show it to you, so if it’s a choice between a dead end residential 30 MPH road and a dual carriageway, and you are plowing on at 70 MPH, it’s going to pick the faster road. It’s constantly doing a fitness test with the GPS and accelerometer data, and it’s model for how you drive.
Then there is the really smart bit. Everyone else who is using Maps, or just an Android phone with location services enabled, are constantly broadcasting their GPS data back to Google’s servers. That’s how it gets the live traffic data, although it’s easy to manipulate as an artist in Berlin demonstrated with 99 phones in a little wagon causing according to Maps a horrendous traffic jam. Anyway, by using other road users’ GPS data and comparing it to yours, Maps can get a better fix on where exactly you are.
But in the end, it’s all just a big guess. Admittedly a guess with a lot of data to back it up, but sometimes it gets it wrong, hence why it can think you are driving on the wrong road, not show you anywhere near your exit when you are practically on top of it, and the various other inaccuracies it displays.
So how do we make it better? Well, the more data Google has the better a guess it can make. Personally I’ve seen it improve greatly over the last 10 years, and while it’s routing capabilities still infuriate me (how many times have you been taken down single track country road when you really didn’t need to, all to save a single minute?), it’s actual tracking is doing pretty well.
If you are having issues with it, you might want to make sure your phone has direct line of sight to the sky as any interference from your car’s roof, your pocket or anything else can make the fix less accurate. You can also make sure your device supports using multiple GPS sources, in my case my Oneplus 7T Pro supports not only GPS, but GLONASS and Galileo among others.
If you need more accuracy, say for tracking your lap times and lines around a race track, or getting accurate 0-60 times, using a device with a higher refresh rate is key. This is a Racebox, a 10Hz GPS receiver meaning it’s 10 times faster than my phone, getting updates every 100 ms rather than once per second. That means the data can be much more accurate especially at high speed as it can plot more points. Here’s a visualisation of that, you are driving round a corner at Donington taking the peak racing line. Your phone’s GPS would report at these dots, see how far they are spaced out? Now you can draw a line through them, but it’s going to include a lot of guesses. Now look at the same corner tracked with the Racebox. Much more data points, meaning you can see exactly where you are with no guesswork involved.
