Sub-Meter Accuracy

Discussion in 'General GPS Discussion' started by deecee, Sep 18, 2003.

  1. deecee

    Carsten Kurz Guest

    You do not receive WAAS, but EGNOS from Switzerland. And if you trust
    the EPE display of your receiver, you're lost.
    Currently, EGNOS does not yet improve any enabled GPSR accuracy.

    - Carsten
     
    Carsten Kurz, Sep 23, 2003
    #41
  2. Interesting... Enable WAAS == enable WAAS option in the handheld. This connects
    to the two available sources (AOR-E = 33 and IOR = 44, I believe) of EGNOS.
    The handheld has one option for the three systems (WAAS, EGNOS and MSAS).

    I have no valid reference to validate the accuracy of the estimation, only
    consecutive measurements from the unit itself. And they look like being
    consistent. Could you explain why EGNOS is not working as it should? As far
    as I know, from April 1st we should have the corrective factor uncertified
    but already enabled...

    Thanks,
     
    Roberto Divia, Sep 23, 2003
    #42
  3. deecee

    Sam Wormley Guest

    Trimble's GeoXT replaces the older GeoExplorer III and is more capable.
    Both feature Carrier phase data collection.
     
    Sam Wormley, Sep 23, 2003
    #43
  4. deecee

    Sam Wormley Guest

    Hi Roberto--Are you referring to measured statistical error, or receiver
    deceiver displayed EPE? Thanks.
    Regards,
    -Sam
     
    Sam Wormley, Sep 23, 2003
    #44
  5. deecee

    Sam Wormley Guest

    WAAS is a regional augmentation of GPS in North America and
    is operational at this time. EGNOS is regional augmentation
    for Europe and is in intial testing stages.

    An Essential Element of EGNOS, like WAAS -- spatial correlation.

    "Differential GPS involves the cooperation of two receivers, one
    that's stationary and another that's roving around making position
    measurements. The stationary receiver is the key. It ties all the
    satellite measurements into a solid local reference." -- Trimble
    Navigation Ltd.

    In the case of EGNOS (and WAAS), the "reference receiver" is a
    network of ground stations. The WAAS differential corrections
    based on that reference network would be valid for a EGNOS (and
    WAAS) enabled GPS receiver that is within or spatially close to
    the reference network. One would not expect valid differential
    corrections from a reference network that exists on another
    continent.

    -Sam Wormley
    http://edu-observatory.org/gps/dgps.html
     
    Sam Wormley, Sep 23, 2003
    #45
  6. deecee

    Ron Wilson Guest


    Actually the GeoXT replaced the GeoExplorer III, which has now been
    discontinued. The GeoXT is more expensive and more accurate.
     
    Ron Wilson, Sep 23, 2003
    #46
  7. deecee

    Ron Wilson Guest


    If you think that WAAS is superior (all things being equal), then why do you
    find it so hard to believe that the GeoXT can give submeter accuracy using
    only WAAS corrections? Is it because the GeoXT is L1 only?


    What software did you use to post-process the data with the GeoExplorer III
    to get 1 cm RMS?
     
    Ron Wilson, Sep 23, 2003
    #47
  8. deecee

    Keith Guest

    Interesting... Enable WAAS == enable WAAS option in the handheld. This connects
    Logging your position with SA Watch will show you that EGNOS doesn't
    improve the reliability (i.e. dispersion is the same and sometimes
    worse than without it)
    ___________________
    Keith
    www.gpspassion.com
    All things GPS and PocketPC
     
    Keith, Sep 23, 2003
    #48
  9. This premise is flawed. WAAS (WADGPS) is almost always better than DGPS
    if the receiver is capable of taking advantage of its features. the idea
    that it is only general is highly flawed. DGPS is also only general
    unless you are setting right next to the beacon tansmitter. Lets look at
    the data a bit. Note that consumer GPS receivers do not take full
    advantage of the WAAS abilities so this discussion does not necessarily
    apply to a consumer unit.

    DGPS does not analyze the incoming data but only generates a single
    correction for pseudo range data based on its location and sends
    corrections for upto 9 SV's max. It does not know the makeup of the
    error components.

    WADGPS breaks down the error sources by removing but not sending
    corrections for tropospheric errors. Understands and sends separate
    corrections for clock errors and ephemeris errors. It sends a correction
    for ionospheric errors that is based on a sophisticated analysis of not
    just one location but all the locations in the vicinity and will be
    highly accurate near a ground station reference but also accurate
    between ground station references by integrating data from a number of
    sources providing information not available to DGPS beacon transmitters.

    Capable receivers can apply their own tropospheric error corrections
    based on their own capability. They can correct for multipath
    separately, they can use the main three error souces in appropriate ways
    that can't even be approximated by the DGPS solution. For example, if
    they have better ionospheric information they can compensate for this
    and still use ephemeris and clock corrections. One such compensation
    could be time of day for example.

    So if you want sub
    If you mean by that you can supply your own beacon transmitter within
    very close proximity then I would agree but if you are talking about the
    availabilty of the government beacon transmitters then see above. Also
    this is not real time by any stretch.
    Also not real time. Kinematic receivers can do this in real time, not
    post processing.

    Dale
     
    Dale DePriest, Sep 23, 2003
    #49
  10. deecee

    Carsten Kurz Guest

    Yes. They should better call it 'SBAS', but since Garmin is a US company
    and WAAS is a US system (and was the first of these SBAS systems in
    operation), they just do not want to irritate their US customer base.
    Enabling WAAS enables WAAS, EGNOS, MSAS (automatic selection of
    satellites).
    EGNOS is not working as it should, because it does not even exist
    currently. What your receiver get's from AOR-E and IOR is ESTB signals.
    You can easily check the quality of the EPE indication by logging the
    NMEA output of the receiver, or create a time-triggered tracklog of a
    static receiver. You will be surprised.

    I have access to two identical receivers and thus can compare
    ESTB/non-ESTB logs of the same timespan. Currently there is NO ESTB
    correction quality visible, only spurious over very short periods of
    time, but not in general. They are still deploying RIMs and improving
    the algorithms. Wait for next year to see real improvements. What they
    recently did was just changing the msg format so that most consumer
    receivers are now able to receive the ESTB test signals. There will be a
    compliant msg format change once the system goes into EGNOS operation.

    Currently it seems that all consumer GPSRs simply seem to 'weight' their
    ordinary EPE figures by some sort of 'WAAS' factor (e.g. dividing by two
    or so, or weighting by the number of corrected satellites). Maybe it's
    not so easy to calculate a more reliable WAAS EPE. Application of SBAS
    corrections is not a 1:1 standard procedure - manufacturers have to
    develop their own correction code and make their own experiences. There
    is a learning curve.

    - Carsten
     
    Carsten Kurz, Sep 23, 2003
    #50
  11. Receiver display and some samples from time to time, nothing statistical.
    Plus some news gathered around the 'net concerning the same brand/model
    of handheld.

    Ciao,
     
    Roberto Divia, Sep 23, 2003
    #51
  12. deecee

    John Bonde Guest

    Submeter didn't surprise me. It was how far under "submeter" you were.
    I just haven't seen any test numbers, other than yours, showing accuracy
    at the 0.6 meter (95%) level. That doesn't mean I think your numbers
    are wrong - just that I'm looking for more data from others. If you take
    your GeoXT anywhere in the WAAS service area can you expect the
    same level of accuracy or will it vary and by how much? I expect
    it will vary given the nature of WAAS. So was your 0.6 meters the
    best you can hope for, about average, or rather poor compared to the
    complete service area? More numbers from others would help
    answer that.
    I typically use Waypoint Consulting's GrafNav/GrafNet software
    as it works well with a variety of manufacturers equipment. Any software
    capable of "fixed-integer" processing will do it.
     
    John Bonde, Sep 23, 2003
    #52
  13. deecee

    stretch Guest

    Thanks! Will do some experimeting myself.
     
    stretch, Sep 24, 2003
    #53
  14. deecee

    Ron Wilson Guest

    John,

    I checked out the GrafNav/GrafNet
    website and it looked great. The only problem is that I couldn't find a
    price anywhere for either of their packages.

    Also, I would like to pick your brain, since you obviously have experience
    doing carrier phase post-processing and I don't! Since I potentially have a
    unit that could acquire 1cm accuracy, I would like to be able to use it for
    that, at least for some applications. Most of what I do, frankly, doesn't
    need that level of accuracy, but there might be some tasks that would
    benefit from it. One that springs to mind is the mapping of wetland flags
    on some of the wetland delineations we do. This would involve only
    capturing a series of points along a woodland or open canopy line. Some
    questions that spring to mind are:

    1) Roughly how much do the various GrafNav/GrafNet software packages go for?
    If it's ridiculously expensive (like $2000 or so),
    then I might reconsider whether or not it's worth it.

    2) How long do you have to average to acquire 1cm accuracy (or close to
    that), especially in a wooded environment?

    3) Which of the GrafNav/GrafNet software packages would I need? I know they
    have a lite version, but it doesn't guarantee 1cm accuracy.

    4) How much of a learning curve will I have to climb to be able to use this
    software with my unit? I read a bit of the online manual, and I was not
    familiar with some of the terms.

    Thanks in advance for your time and knowledge,
     
    Ron Wilson, Sep 25, 2003
    #54
  15. deecee

    Carsten Kurz Guest

    I only have a german pricelist, it may be more expensive over here:

    GrafNav/GarfNet (kinematic) - about 5000US$
    static only - 3500US$
    Lite - 2500US$

    So, it is ridiculously expensive - but it also seems ridiculously
    capable.

    Does anyone know how much the Trimble/Terrasat software is?

    If you only want to go for static single points, maybe some of the
    online services will also do. GrafNet and similiar packages will
    certainly allow more analysis and a better insight in positioning
    reliability.

    - Carsten
     
    Carsten Kurz, Sep 25, 2003
    #55
  16. deecee

    John Bonde Guest

    The software is ridiculously expensive. The numbers Carsten posted
    seem about right. Software costs have always been the biggest part
    of cm-level GPS costs. The hardware can be quite cheap with OEM
    boards going for as little as ~ $100US.

    There is a significant learning curve for getting comfortable with
    carrier phase processing. Like anything it depends on a host of
    variables, but you're not going to get really proficient in days or
    even weeks.

    Remember that there are many more limitations to collecting
    data for cm-level processing, especially L1-only processing. The
    basic requirement for ~ 1cm accuracy is to keep carrier lock on at
    least 5 satellites for ~ 10 minutes. Once that is accomplished a
    good software package can usually "fix the integers" and give
    you cm-level accuracy. If the software is sophisticated enough
    it can do this even when the receiver is moving. Then, as long
    as you never drop below 4 satellites you maintain the cm-level
    accuracy. If you do drop below 4 sats the initialization
    process needs to be restarted. Note that when post-processing
    you can process the data backwards in time so if you dropped
    below 4 sats for only a short period in the middle of an hour
    long session you might only loose cm-level accuracy for that
    short period.
    The above description is approximate. I can't go into every
    detail here, but the point is that since you need 10+ minutes of
    continuous carrier lock for cm-level accuracy it's not going to
    happen while walking around under even fairly light canopy.
    It is possible for static sessions under light canopy though.

    One last point on carrier phase processing. There is what's
    called a "float" solution and a "fixed" solution. Most processing
    packages start off with a float solution. Here the accuracy
    starts at meter-level and works its way down to a few cm
    over time. The time period for this to happen can be
    anywhere from 10 minutes to an hour or more. This is
    all some packages provide so while they can provide
    accuracy at the few cm level you may not want to wait that
    long. It's also why you see specs like 30cm for your receiver
    with carrier phase processing. Fixed solutions make some
    guesses, and, if correct, accuracy goes immediately to
    cm-level.

    A package like GrafNav does make learning post-processing
    easier, but since you don't really need cm-level accuracy you
    may not want to spend several thousand dollars. I'm guessing
    that you would just like to see how accurate your receiver can
    be so the thing to do would be to look for some cheaper
    software. What you might find will certainly be more limiting,
    but it might get you a start.

    Terrasat use to provide a free "float" version of their software.
    They were bought out by Trimble and I'm not sure if the offer
    still stands. It's worth checking out.

    Ashtech's old MSTAR program does provide both a float
    and fixed solution for static baselines. You can download it.
    ftp://ftp.thalesnavigation.com/pub/software/mstar/MSTARsoftware/

    It works with RINEX data. The carrier phase processing
    isn't state of the art, but it's something you can play with. In
    general it needs longer time periods and it may need fast
    data collection rates. Something like at least every three
    seconds. I don't remember exactly. NGS's INTERPO.EXE
    program available on their web site can come in handy
    here. Interpolating carrier phase data can be tricky though.

    There may be other relatively inexpensive programs, but they
    will probably be "float" only and also probably static only.
     
    John Bonde, Sep 25, 2003
    #56
  17. deecee

    Ron Wilson Guest

    Carsten,

    Thanks for the prices. I'm afraid that's a little more than I had in mind
    for what I'm going to do with post-processing.
     
    Ron Wilson, Sep 26, 2003
    #57
  18. deecee

    Carsten Kurz Guest

    There are I guess 4-5 free or low cost packages that should do nicely at
    least for static applications. There are also some internet based
    services (but some of them need dual frequency rover data).

    - Carsten
     
    Carsten Kurz, Sep 26, 2003
    #58
  19. deecee

    Ron Wilson Guest

    John,

    Thanks a million for all of your help and suggestions. You're right about
    the ridiculously expensive part, but businesses that must have that kind of
    accuracy will pay that much without batting an eye .
    My sessions would be static.
    Well, it sounds like I need to do some more homework, before leaping into
    this facet of GPS. Once again, thanks for all of your tips and suggestions.
    I think I will start with the least expensive solutions, just to see if I
    get any significant improvement and then gradually work my way into more
    rigorous treatments.
     
    Ron Wilson, Sep 26, 2003
    #59
  20. deecee

    Andy Guest

    You get 1cm rms on a Geo III? I'm impressed. And I'm fairly sure the guy
    who designed it will be when I tell him ;)

    A general note on the way the XT accuracy is specified: It's a mixture
    of the theoretical accuracy and a large amount of measurment and
    testing. Rather oddly the two do tie up fairly well. Typically a unit is
    left on the roof of the building or outside somewhere with fairly clear
    skies and run terrasync. A position is logged every second for either 24
    hours or until the battery is flat (about 10 hours). The standard
    deviation of the points logged is then calculated after the appropriate
    type of post processing.
    So a claim of around 30cm carrier corrected accuracy means around 66% of
    all locations will be within 30cm of the average. I'm not sure if this
    is used to calculate the specs or just to check the calculations are
    correct but it is what the specification means in the real world.

    It's not all purely an attempt to sell more expensive recievers, honest.

    Actually a trimble survey reciever downgraded to the capabilities the XT
    (L1 only non-RTK etc..) and the Geo XT do give very similar results,
    they are afterall similar basic designs. The survey reciever is however
    very slightly better (typically ~2-3% lower SD), there is just more
    space to spread the electronics out and so there is slightly less noise
    from the recievers processor interfering with the GPS.
    With the survey recievers you're also paying a lot for the RTK features,
    even discounting the extra hardware required that's a lot of
    development on a relativly low volume product, they have to make the
    cost back somehow.


    ____________________________________________
    reply address : newspost at . org
    talon . uk
    ____________________________________________
    I'm not paranoid about spam.
    Well ok, maybe a little bit.
     
    Andy, Sep 26, 2003
    #60
Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.