What is GPS 3? GPS III vs GALILEO and GLONASS – which is best?

GPS III Explained – What is GPS III?

What is GPS III (GPS 3)?

GPS III is America’s upgraded GPS system which will perform with a similar level of accuracy to Europe’s GALILEO. It is used to enable electronic devices to tell you where you are.

Image: Lockheed

As of 2019/2020 the term “GPS” specifically refers to the current GPS II (GPS 2) system of American navigational satellites. We all use the term ‘GPS’ incorrectly in that sense. GLONASS is Russia’s version and GALILEO is Europe’s version. The correct term that we should all use is GNSS rather than GPS. GNSS stands for Global Navigation Satellite System. GLONASS is a GNSS, GPS is a GNSS and GALILEO is a GNSS. There are other regional systems too.

These systems can be used on their own but increasingly they can be combined.

Thus GPS+GLONASS usage is now fairly widespread in sports devices and GPS+GALILEO is now increasingly used in high-end SMART devices in phones and sports watches, including the new Garmin Fenix 6.

Let’s first talk a LITTLE bit about ACURACY, GLONASS and GALILEO as that then better sets the picture to explain where GPS III will fit. I’m also going to tbe talking about sports smartwatches but the exact same principle applies to car satnavs, smartphones and other handheld navigational devices.

NB: I’m using the term GPS to mean GPS II from now on…


There are many factors affecting the positional accuracy you will experience with your device on the ground.

I want you to put in the forefront of your mind two factors, namely the NUMBER of satellites and the transmission frequencies the satellites are using. It seems obvious that the number of satellites you use will increase accuracy (it’s not strictly true though) and the other point to bear in mind is that if two frequencies of positioning signals are used then there is a bit of physics that will mean that you will be able to reduce accuracy losses from the route the signal takes (ionospheric delay) to reach your device from the satellite.

Detail on accuracy eg looking at HDOP, PDOP

GLONASS Accuracy

GLONASS is neither inherently better nor worse than GPS (GPS II)

HOWEVER, when GLONASS and GPS are used TOGETHER the likelihood of accuracy IS increased. Simply because there are more satellites available in both ‘constellations’ of satellites and hence a higher chance that your watch/smartphone will be able to see enough of them for a good positional fix.

It is said that using GPS+GLONASS can give accuracy down to 4.5/5m and can also improve accuracy in built-up areas; especially in the Northern Hemisphere. That is THE SAME level of accuracy as standalone GPS. It’s just that the LIKELIHOOD of achieving it MAY be increased.

GPS+GLONASS is NOT more accurate than GPS alone – I have consistently found that in my product testing here

MORE GLONASS details (here)


No. The OTHER Galileo

As of Q4.2019 Europe’s GALILEO constellation is nearly complete with the final 2 (of 30) satellites expected to be operational in 2020. However, 24 satellites are all that is required for the system to be fully live and, as far as I know, that is the situation now.

The same principle applies with using GALILEO satellites alongside the GPS ones as it did for GLONASS…the more the merrier.

HOWEVER, we have the potential for AWESOME levels of accuracy to better than +/-1m, even in cities, because of the use of dual frequencies. GALILEO has 3 frequencies and two of them, E1 and E5a, can be used together with the L1 and L5 frequencies from GPS.

Unfortunately, in order to achieve that, a special chip is required such as the Broadcom BCM47755. And whilst that chip IS eventually intended to be used in wearables, I don’t know of any wearables that use it right now. It is used in some smartphones such as the Xiaomi Mi 9.

The current situation right now with wearables is that the Sony CXD5603GFchip seems to be the de facto standard for GPS and GALILEO signals received by Garmin, Coros, Polar, Suunto and others. Whilst this Sony chip is very low-powered and GALILEO-enabled it does NOT support dual frequencies AFAIK. Consequently, the levels of accuracy we see are largely unchanged from the GPS chips in the wearables of 5 years ago…although the power consumption is dramatically lower and it is power consumption and battery life that is driving wearable tech usage NOT ACCURACY. 🙁

MORE GALILEO details (here)

SBAS and Accuracy

SBAS? Where did that come from?

I threw that one in. SBAS is a mildly important factor that might further improve GPS accuracy which can be enhanced if corrective readings are available to be taken from ground-based transmitters. This is performing a similar task to a dual-frequency signal from satellites but instead from the ground. The availability of these ground-based signals will be less than that from satellites.

Polar is using SBAS with the Sony CXD5603GF chip of April 2019 and is supporting GALILEO from October 2019.

What is GPS III (GPS 3)?

Originally planned for 2014, the next generation of American GPS satellites, built by Lockheed, is finally being launched into orbit. One has already been launched by SpaceX and one last month by ULA. 8 more are planned with SpaceX sending up the next of those in January 2020. These should all be operational in 2023.

Beyond 2026 will see a further 22 GPS3F satellites take to the sky, culminating with the last launch around 2034.

GPS III has been designed to use the same mid-point frequencies as already used by GPS and GALILEO. Thus, from a consumer device standpoint, we should effectively just see this as an expansion of the existing GPS and GALILEO systems ie the availability of even more satellites. eg I’m hoping that the aforementioned Sony GNSS chip will ‘just work’ with GPS III

But these new GPS III satellites also open up a new civilian frequency – L2C.

Dual-Band GNSS

Edit: Added/changed after original text due to comments below, I wasn’t planning to talk about dual-band GNSS too much

The following chart attempts to show that Galileo E1 and GPS L1 can be used together as ‘one band’ and that Galileo E5a and GPS L5 can similarly be used together or individually as another band. By inference E1+L1 and E5a+L5 and L2c represent 3 bands that give us more scope for better accuracy derived from dual-frequency signals

1.559–1.592 GHz (E1)1.593–1.610 GHz (G1)1.563–1.587 GHz (L1)1575.42 MHz L1 (Civilian)
1.237–1.254 GHz (G2)1.215–1.2396 GHz (L2)1227.6 MHz L2C (Civilian)
1.189–1.214 GHz (G3)
1.164–1.215 GHz (E5a/b)1.164–1.189 GHz (L5)1176.45 MHz L5 (Civilian)
1.260–1.300 GHz (E6)

From the comments below there are more smartphone chips than I thought that are dual-frequency GNSS ready/enabled. As well as the BCM47755 chip that I specifically mentioned as its specs claim readiness for WEARABLES, there is also the Snapdragon 855 (from 2018), the Huawei Kirin 980 (from 2018) and Huawei Kirin 990 (from Sep 2019). I’m assuming these are dual-frequency GPS+GALILEO (not GPS III) …so this section is going off at a bit of a tangent to the original intention.

If anyone can send me some dual frequency workout tracks GPX/TCX from a smartphone and a separate track(s) from a sports watch on the same RUN/Ride…I’d love to put up some comparative images here.


GPS III will bring more satellites into view, enabling many devices to stand a better chance of achieving current levels of GNSS accuracy ie +/-5m

When combined with a suitable consumer GNSS chip that can handle multiple frequencies, the opportunity to receive high precision readings will exist.

Opinion? – Contain your excitement for a year or two more. Sports-GNSS usage needs to see the adoption of a more capable GNSS chip beyond the Sony chip used widely in 2019. It’s more likely that we will first see this innovation in smartphones, with sports devices a year or so behind. Finally, let’s not get too hung up on the satellite signals, we see BETTER RESULTS FROM 5 YEARS ago with GPS-only sports watches, clearly all sports watch manufacturers need to look closely at, for example, the design of their antennae.

For Mirko 😉


Reader-Powered Content

This content is not sponsored. It’s mostly me behind the labour of love which is this site and I appreciate everyone who follows, subscribes or Buys Me A Coffee ❤️ Alternatively please buy the reviewed product from my partners. Thank you! FTC: Affiliate Disclosure: Links pay commission. As an Amazon Associate, I earn from qualifying purchases.

37 thoughts on “What is GPS 3? GPS III vs GALILEO and GLONASS – which is best?

  1. Well written and summed up. But how much can the accuracy actually improve? I thought there was a limit to how accurate the GNSS could be, because it’s restricted by the military.

    1. #ItsComplicated (of course)

      there are other aspects of interest to the military such as restricting the ability of others to jam signals.

      as no-one has seen what dual frequency signals can do on a sports watch then all i can say is that i’m hopeful that accuracy will come. and i mean +/-1m rather than +/-5m
      IIRC the military can get EVEN more accurate. but centimetre level accuracy is of no interest to me…+/-1m is

    2. If you want the accuracy to improve you need to start by silencing all of the reviewers, it’s not a technology issue.
      Reviewers don’t really have much useful to say about devices (no, really, read a review more closely!). They can’t be negative or they wouldn’t be a reviewer very long. This means any thorough look at a feature may (often does, trust me!) lead to finding flaws in how it works. This leads to reviews becoming a slightly more in depth version of the spec sheet, often with a focus on battery life because that’s something easy to measure that we can all relate to.
      10 years of reviews talking endlessly about battery life and here we are – 100 hour battery on a device that almost every single user will use for 4 hours maximum. And shitty GPS tracks. Backtrack a few years and your device would hit 8 hours is you were really very nice to it indeed. But it knew exactly where it was.
      The addition of Gallileo etc. is a distraction from the simple truth: Garmin made a trade off based on what they thought people wanted. And most people backed that up by buying every new device they were offered.

      1. 2014/15 was the adidas micoach with 4 hours of gps+music and 8 hours without and the first apple watch, IIRC, was less than 24 hours battery life full stop. from that starting point, the situation then was not acceptable. Clearly the focus on battery was important by the tech companies

        HOWEVER I fully accept that where we are now should mean that battery is not an issue. eg In last week’s Fenix 6 review i used the phrase “PEAK BATTERY”, perhaps battery life cannot now get materially better and certainly doens’t need to get better. so i agree with you.

        i’ve done some polls over the years and GPS accuracy DOES come up at the top of the lists. so it is important to a vocal group of users…minority?

        I still find it amazing how NOT ONE SINGLE SPORTS WATCH CAN ACCURATELY TELL YOU HOW FAST YOU ARE RUNNING. you NEED a footpod for that. PLUS I am ENTIRELY CERTAIN that the AMBIT 3 (2014) and POLAR V800 (2014) are the most accurate consumer GPS sports watches ever produced…as you imply, accuracy didn’t seem to do Polar and Suunto any favours!

      2. “They can’t be negative or they wouldn’t be a reviewer very long.”

        I don’t think that’s true at all. And in fact, I think the opposite. Reviewers who aren’t negative aren’t reviewers very long at all. Everything is a fluff bubble and eventually people see through that.

        I think both 5K and myself can be pretty negative when warranted, one only need to look at my review today to see that. I also think a lot of people don’t actually read the nuance of what either of us are writing. While some of it is talking through how something works, it’s also the details of how it’s not working in a specific ways.

        Much of what I write is preemptively answering questions as I go along through the post. In my semi-ideal world I’d have managed to answer every possible question by the end of the review, mainly ones not covered by the manual.

        I also don’t think either of us actually talk about battery life all that much either. I think for the most part for most people (Apple Watch aside), it’s more than adequate for everyone by those competing at UTMB, and has been for years.

        Though, I disagree with 5K on the footpod thing. I have no issues pacing with GPS today (or for many years). 🙂

      3. I absolutely agree. Why then so much focus by the manufacturers on battery life in the new watches? The Fenix 6’s battery life figures and Power Manager are given some prominence, for example, on the Garmin website. And the numbers are certainly impressive. Now imagine if Power Manager allowed me to trade off a bit of that battery life for better GPS performance.

      4. the cynic in me would say that they do. Just DISable GALILEO and GLONASS 😉

        but i take your point and have thought the exact same thing myself. I’m not sure if a bit more juice can be translated in to slightly better accuracy tho. and as i think i said elsewhere, i’ve given up and use footpods/chest straps

      5. In the era of the Sony chipset, saving GPS power is the low hanging fruit. I suspect music playback is what’s driving it. Music playback has significantly increased the power demands of these wearables and until /those/ chipsets get more efficient the watch makers are going to take their power savings where they can.

      6. some of that is going to be Bluetooth as well in transmitting to the earbuds
        I understand that it SAVES power by using oHR rather than a chest strap. it sounds wrong but the BLE power saving is why oHR is better (for battery life)

    3. There is no military limit on accuracy with GPS except in specific locations during specific military operations (think Iraq and Afghanistan). Selective Availability (intentional positional degradation) was turned off in the Clinton era.

    4. Accuracy used to be restricted with a dithering effect that could add large errors called Selective Availability.
      This was removed some time ago. Accuracy now should be able to be within 1M and perhaps less with the dual frequency systems (L5 signals) and or SBAS space based augmentation system (satellite based differential GPS)

      1. thank you
        i never knew about the selective availability

        i think the problem in ‘accuracy’ found by sporty people (my area of ‘expertise’) is also highly dependent on the design of the receiver. ie failure to achieve the 1m accuracy that is possible with better equipment

    1. thank you, yes I appreciate that it’s not simply a case of switching over gnss chips and turning on a new frequency and that antenna re-design is required.

      someone needs to get re-designing! it’s probably their job 😉

      I would be interested to know, generically, what has happened to antenna design. clearly over the last 5 years something has been compromised in their design within sports watches (I assume space/size)

      1. I think that is because noone really cares about a good track expect perhaps us gps freaks (if i am allowed to refer to us like so)

        Most modern consumer needs a good pace and a consistent gps track.

        For example on a 42km race its difficult to zoom and evaluate the whole run and few will do it, provided that pace and distnace match.

        That said the compromise is design based decision imo.

      2. agreed.
        They do already get the consistent GPS tracks. The majority of tracks are good enough (IMO) for a pretty, post-workout view of where you’ve been.

        But the ‘good’ pace is a thing you just don’t get consistently acorss a wide variety of reasonable running scenarios. The Fenix 6 I’m currently running with is simply wrong and unactionable when it comes to instant pace eg massive 30secs/km variations from a known, constant pace.

        In my tests I find most devices get the total distance to within +/-0.5%….which sounds pretty good. But that includes under-counting and over-counting of multiple incorrect segments of the run. +/-0.5% would not translate to 26miles+/-0.5% in an urban marathon I would wager. The error would be larger because of the buildings (urban canyon effect)

        freaks: I might have once counted myself as that but I realised some time ago there was just no point in obsessing about it. #Stryd.

      3. Meh, my Stryd still routinely underreports distance by at least a couple of % and when I look at my runs in PowerCenter vs. Movescount/Strava, PC reports an average pace ten to fifteen seconds slower than MC. Just a heads-up that Stryd explicitly don’t guarantee that their product is accurate for everyone and if you’re one of the unlucky ones (like me), the only way to get accurate instant pace EVEN WITH STRYD is to turn on calibration in the watch.

        I don’t trust the power figures either because it turns out that PC also massively underreports ascent. That being the case, there isn’t much data in PC that’s actually useful to me and instant pace is the only useful function I’m getting out of it (subject to autocalibrate being turned on in the watch).

        Which is OK, up to a point, because instant pace is the primary thing I wanted it for… but it also means the footpod is way overpriced in my book.

    2. Bringing up antenna design is critical here.

      Dual frequency to solve ionospheric delay will help with absolute accuracy. However that is not what plagues wearables. Wearables suffer from unaimed antennas which therefore have no (or minimal) ground planes.

      An antenna which is always oriented up can have a ground plane to greatly mitigate reflected signals entering the antenna from below. A wearable can not. This means wearables are highly susceptible to multipath errors and these errors are (when the antenna is moving) largely random in nature – creating ugly tracks. Ionospheric delays are largely non-random in nature and will tend to cause systematic shifts in positions, offsets, not noise.

      Signal modulation can help mitigate multipath errors, and Galileo is far better than GPS or GLONASS on this front… at least in theory. However on the wearable front we’re not actually seeing it (yet?).

      1. I agree on this. If you try to wear your GPS watch clipped to your cap or something facing upwards all the time. GPS track will be superb. When you run, your watch on the wrist keep on moving and if you are in the city or where there is a lot of obstructions like tall buildings, bridge, trees ..etc.. your GPS path tend to turn ugly. This is because there are many multipath errors.
        Probably having a separate external GPS antenna that can be clipped to your shirt or cap for better orientation and connected via ant+/BLE to the watch might be a better solution to those looking for better GPS accuracy.

  2. I still don’t really understand why GPS performance isn’t a priority for watch manufacturers.

    The one thing that everyone knew about the Polar V800 is that it had stellar GPS accuracy. It was a great selling point for the watch – perhaps the only one as it became surpassed in ever other aspect by the competition. I just assumed that Polar would do everything they could to make sure the Vantage matched the V800’s accuracy. Perhaps they tried to?

    Garmin must spend huge resources coming up with new features we never knew we wanted. They surely must also spend time worrying about each new watch’s GPS performance. If the antenna is so important, their engineers are no doubt continually improving the antenna design in each new watch.

    People tend to say that GPS on sports watches is good enough. So I am always a bit surprised at just how wonky my GPS tracks sometimes look when I run in a city or in a valley. It can’t just be down to my excessive arm-swing.

    I think I must have misunderstood something along the way.

    1. polar specifically and corretly (IMO) targetted excellent sensor performance at the launch of Vantage. but…
      next month they will have SBAS and GALILEO in the Vantage. Potentially the best ever setup for GNSS accuracy. #LetsSee
      yes Garmin’s R&D budget is listed in their accounts and is impressive. Innovate or die. They innovate a lot.
      yes you would have thought that garmin ‘continually improve antenna design’ but…here we are they are still ‘meh’

      the earlier comments from DaveLusty get to the heart of the matter. THE KEY thing for the last few years has been battery life. Mayeb one compromise was smaller space for an antenna in favour of a slightly larger battery…that kind of thing. There’s no point having the most accurate device ever coupled with a continually flat battery. maybe from now on we will see EVEN more of a focus on sensor accuracy?

      “gps on sports watches are good enough”…but for what purpose? I always say it’s good enough for a post workout pretty track. but it’s rarely good enough for instant pace. it’s usually good enough for navigation. different people with different needs say different things

  3. Nice article. Hopefully the expert of GPS of the DCRainmaker forum, “Dom”, could chime in and give us some other news or scoop about some new gps chipset and their adoption in the consumer market.
    DCR also noted that Garmin is aware of the advantages of the dual frequency gps, but that now is too soon to release something in the market. The dual frequency chipset will appear in watches only after they are mainstream in the phone market. And, after all, the advantages of dual frequency in watches that have so little battery and a little antenna have yet to be verified.
    I solved the problem of instant pace with the footpod Runscribe (I bought it when I realized that brands were not going to install immediately dual frequency chipset in watches) and now I don’t need any more the accuracy that at first I was waiting from dual frequency gps.
    I’m running now with a FR945 (that has the Sony chipset) and instant pace of gps is good in open areas where the signal has no obstacles (easy for gps), but it’s enough to introduce some difficulties (trees, building or mountains) and I’m seeing what you reported (sometimes instant pace can be 30 seconds off of real pace).
    Instant pace of the FR945, when not connected with a footpod, as noted in the Garmin forums, is also very smoothed and so it’s good for steady runs, but not so good when there are frequent changes of speed. In the interval screen of the structured workout Garmin uses lap/step pace, that is much faster at the beginning of the interval to react in changes of speed than instant pace, maybe because at the beginning of the lap it uses the data of accelerometer and not the data of gps.

    1. “gps is good in open areas” …exactly so.
      “hen not connected with a footpod, as noted in the Garmin forums, is also very smoothed and so it’s good for steady runs, but not so good when there are frequent changes of speed.” I’m getting poor pace onthe F6 with a constantish pace for 90 minutes.

      who is the mysterious Dom?

  4. MI9 has no that BCM47755 chip, MI9 has Snapdragon 855 cpu which has inbuilt dual band GPS already (all mobile with 855 has such a feature btw), MI8 has Snapdragon 845 only, which has no such a feature , so they had to put that Broadcom chip next to it and this is the only one mobile which has this BCM chip.

    The Kirin980 and 990 cpus also has dual band GPS chip (Huawei)

    And there is a special GPS chipset, AllyStar8040, which knows the most bands (in mobile) on 136channel, especially knows the “L5” signal from Beidou (the chinese GNSS), and only 1 mobile has it, the cheap Lenovo K10 Note (or Z6 Youth/Lite in chinese market, same product just different name):
    GPS (US): L1, L1C, L2C, L5,
    QZZS (JP): L1, L1C, L2C, L5,
    GLONASS (RU): L1, L2
    GALILEO (EU): E1, E5, E6
    BEIDOU (CN): B1C, B1I, B2A, B2I, B3I
    IRNSS (IN) : L5
    And all SBAS : WAAS , EGNOS, MSAS , GAGAN, SDCM and these have more band also 🙂

  5. just FYI: a Huawei P30 mobile in cycling jersey backpocket has better track in wooden mountains (where the signals are the worst) during a ride than every other bike computers or watches from that big group (and everytime, not just a single day). On front mount, it is perfect.

    And yes, the good antenne (design) is the most important at mobiles (too) for perfect reception, the chip is secondary.

    1. thank you for all that info
      I would love to see the tracks you talk about.

      I assume there is much more space in a smartphone for good anetennae
      As per https://www.qualcomm.com/products/snapdragon-855-plus-mobile-platform , Snapdragon 855 (Systems: GPS, GLONASS, Beidou, Galileo, QZSS, SBAS, Dual frequency GNSS) is in all of these

      Asus RoG Phone 2
      Asus Zenfone 6Z
      Blackshark 2 SKywalker
      Galaxy Note 10
      Garmin Fenix 7 (just kidding…I added that one in myself 😉 )
      LG G8 ThinQ
      LG V50 ThinQ
      OnePlus 7
      OnePlus 7 Pro
      OPPO Reno
      Samsung Galaxy S10
      Sony Xperia 1
      Xiaomi Mi 9
      Xiaomi Mi 9T Pro
      Xiaomi Mi MIX 3
      Xiaomi Redmi K20 Pro
      ZTE Axon 10 Pro
      ZTE Nubia Red Magic 3

      1. i will search them on the week and share soon.

        The problem is that this AllyStar (or any other chip) has 5x more power consuption than Sony chip, the operation voltage also much higher, its size also bigger a little bit , so i doubt we will see these in F7, the Sony (or others) must develop a new chip with dual band capability for wearables…

      2. “the Sony (or others) must develop a new chip with dual band capability for wearables…” I agree with this, because I believe that Sony is the only brand now that is producing chipset with a different technology that is especially good for low energy consumption. I’m curious too to know if Sony is working on a chip with dual frequency capabilities, it could be the chip inside the Fenix 7.
        Probably Garmin will continue to install Sony chipset for some more years.
        In the past years, Garmin used for some years initially Sirfstar and then Mediatek, and it never changed frequently chipset brand.

  6. “the5krunner
    September 17, 2019 at 10:16 pm

    I know I know !! We tried to sort you out. But failed 🙁 You must be one of the tiny subset of people for whom its pants 🙁 🙁 two sad faces…sorry”

    (For some reason there isn’t a “reply” link to this post up above, so I can’t see any alternative but to start a new thread)

    No worries, I’m not whining at YOU, I just want to put my experience out there for others to hear about and make a more informed choice. Those of us for whom it doesn’t work may not be as tiny a subset as you might think, given that I see quite a few mentions of calibration factors on the Stryd discussion forum. But I only found that forum after I’d purchased and discovered that the much-vaunted accuracy-to-the-metre wasn’t happening for me (and honestly, I don’t think I’m a particularly weird runner). That was a severe disappointment and I probably wouldn’t have dropped the price of a Spartan Trainer watch on the footpod if I’d known in advance.

    (I do still use it — with calibration turned on — because I’m hopeless at gauging my own pace while running. On the plus side, Suunto’s autocalibration seems to be pretty good at seeing through the footpod’s shenanigans; it usually underreports distance, is occasionally spot-on, and rarely overreports by an outrageous margin, and the watch manages to compensate for all variants and give me plausible distance data in Movescount.)

  7. Runners and cyclist usually do not run or ride alone. What about pooling multiple devices using Wifi to assist each other? Assisted GPS?

    1. hmmm. interesting idea, i hadn’t thought of that.
      other ground readings are how SBAS works but the base station there would be fixed, i’m not sure how several moving bodies (all in different positions) would help. definitly a case for someone good at maths to come up with a good algo.

      1. Triangulation using Time Distance of Arrival might work. But yes it would require a good algo that knows when GPS signals are obstructed such as under trees. Basically a device with good signal needs to assist a device with obstructed signal. The privacy policy will also be interesting.

        SBAS looks promising but seems to go by different names in different parts of the world. Hopefully they compatible with each other.

Comments are closed.