search

SignalK / specification

Signal K is a JSON-based format for storing and sharing marine data from different sources (e.g. nmea 0183, 2000, seatalk, etc)
Other
91 stars 69 forks source link

Satellites in view #163

Open tkurki opened 8 years ago

tkurki commented 8 years ago

Extend schema to support satellites in view information as available in GSV and PGN 129540

tkurki commented 8 years ago

https://github.com/SignalK/n2k-signalk/issues/12

faceless2 commented 8 years ago

I'll start the ball rolling then:

"gnss": {
    "satellites": {
        "constellation": {
            "gps": {
                "1": {
                    "elevation": 85,
                    "azimuth": 24,
                    "distance": /*not sure this is useful */,
                    "snr": -76,
                    "active": true      /* whether the satellite is being used for a fix */
                },
                /* keys continue, key being the PRN of each satellite */
            },
            /* keys continue, being other constellations of satellite eg GLONASS */
        }
    },
}

Justification:

  1. Keying the satellites by PRN makes updating the model easier, and no arrays are required.
  2. An "active" field means you can combine the GSV and GSA NMEA0183 messages into one - these are all the satellites in view, and some of them are also in use (active).
  3. I'm not familiar with the N2K PGN referenced so would need clarification on the meaning of "Range Residuals" is and what values of "Status" are other than "Used". If it's just "Used or "Not Used' then the "active" field would cover it.
tkurki commented 8 years ago

Is there something that should appear as a sibling of constellation? Why should it be there?

Other than that I'm good with this - can you take it further?

I have no in depth knowledge of the 129540 details either.

faceless2 commented 8 years ago

You're correct - either "constellation" or "satellites" is redundant and could be removed. "satellites" is more immediately intuitive, all else being equal I'd suggest we keep that one.

By "take it further" you're going to make me learn to use Git and submit a pull request aren't you? Argh. I knew this day would come, but I have been putting it off :-) Will try to get up to speed on this and on the spec structure.

sumps commented 8 years ago

Its easier than you think, I had never used it and when I met the guys at METs I promised I would have a go. Once I sat down and started, I found the GitHub part was easy and the only thing holding me back was my knowledge (zero) of JSON. You do not need to clone a repository locally on your PC, just start by going to one of the files in the Master branch that you want to change and once you are looking at the contents of the file just click on the Edit (pencil) icon. This allows you to immediately start editing the file and automatically creates an auto-named branch of Master with just the file you are editing. Once you have done the change, you click to create a PR, add your notes and justification for the change and post it. Then others will review and merge it for you. Go on give it a go ! image

faceless2 commented 8 years ago

An update based on comments above and also some data derived from @timmathews original at http://signalk.org/api/gnss-field-guide.html which I've just seen.

"gnss": {
    "satellites": {
        "gps": {
            "1": {
                "elevation": 85, /*elevation in degrees */
                "azimuth": 24,  /* azimuth in degrees */
                "range": 123123123, /* distance in meters from centre of ellipsoid, optional */
                "snr": -76,
                "active": true      /* whether the satellite is being used for a fix */
                "differential": false   /* whether fix on this satellite is differential. default is false */
            },
            /* keys continue, key being the PRN of each satellite */
        },
        /* keys continue, being other constellations of satellite eg GLONASS */
    },
}

Changes were adding range in meters - definitely useful as with elevation/azimuth it allows us to convert between other representations of position, e.g. ECEF. And adding "differential", for a differential fix. This field plus "active" would allow us to represent all the options from PGN 129540:9..+7 - well, except for "Not tracked" which I don't really understand. There is no analogy for this status in NMEA0183.

Note this doesn't (yet) cover satellites reported by multiple sources: if we have two GPS units, each will have a different set of satellites. Do we report this? If so, how? Needs more thought.

keesverruijt commented 8 years ago

elevation and azimuth need to be in radians?

SNR is a ratio, but then we lose the logarithmic nature so should we maybe add a dB unit?

tkurki commented 3 years ago

Need to account for

Today I would rather start from deltas than full, as that maps the real world incoming data better to message granularity.

Also need to consider linking a position update to the underlying fix data. Is this just by coincidental timestamps or is some other mechanism possible?

tkurki commented 3 years ago

A log snippet with multiple GSVs and GSAs

$GNGGA,111103.500,5054.3949,N,00107.7952,W,1,12,0.5,50.8,M,46.7,M,,0000*64
$GPGSV,3,1,09,07,16,321,37,08,29,281,33,10,29,143,35,16,75,216,35,0*6E
$GPGSV,3,2,09,18,38,057,35,20,44,105,40,21,81,117,33,26,43,164,25,0*63
$GPGSV,3,3,09,27,62,289,41,0*5B
$GLGSV,3,1,10,65,14,112,18,71,15,018,11,72,25,069,31,77,10,181,30,0*79
$GLGSV,3,2,10,78,52,221,38,79,44,310,28,80,00,342,,81,35,261,40,0*7E
$GLGSV,3,3,10,87,41,052,31,88,75,350,33,0*73
$GAGSV,2,1,07,01,37,308,33,03,09,074,35,05,07,025,,13,85,237,31,0*7F
$GAGSV,2,2,07,15,39,060,33,21,63,228,39,26,30,239,40,0*44
$GNGLL,5054.3949,N,00107.7952,W,111103.500,A,A*5F
$GNZDA,111103.500,08,07,2020,,*41
$GNRMC,111103.600,A,5054.3949,N,00107.7952,W,0.02,0.00,080720,,,A,V*1E
$GNVTG,0.00,T,,M,0.02,N,0.0,K,A*11
$GNGGA,111103.600,5054.3949,N,00107.7952,W,1,12,0.5,50.8,M,46.7,M,,0000*67
$GNGSA,A,3,18,27,16,20,21,08,26,10,07,,,,0.9,0.5,0.8,1*35
$GNGSA,A,3,78,81,87,79,65,88,71,72,77,,,,0.9,0.5,0.8,2*31
$GNGSA,A,3,03,21,26,01,15,13,,,,,,,0.9,0.5,0.8,3*34
tkurki commented 3 years ago

Note the talker ID to distinguish between the systems:

https://www.u-blox.com/en/docs/UBX-13003221 image

Hakansv commented 3 years ago

@tkurki Although a JacaScript novice I've tried to add the satellite constellations. My first attempt gives the result as of the pict below. See also (mostly up) http://signalk.stupan.se:3000/signalk/v1/api/vessels/self/navigation/gnss The code is here: https://github.com/Hakansv/nmea0183-signalk/blob/addGSV/hooks/GSV.js I divided the constellation path. One for each GSV message. I can't say I'm fully satisfied with the outlook. Any advise to code in another way is appreciated. Like the "satelliteConst_msg1" But so far all data needed for OCPN are there and would be possible to catch but as mentioned above there is more in life than O. (?) Thanks Håkan

bild

Hakansv commented 3 years ago

Another way to name the constellation path: Views? Håkan

navigation:
    gnss:
        satellites:
            value:                  10
        satConstellation:part:1
            value:
                satNr:                  "2, 26, 29, 7"
                elevation:              "0.7165, 0.2618, 0.2259, 0.8356"
                azimuth:                "4.756, 0.5061, 0.2567, 2.735"
                SNR:                    "29, 31, 35, 24"
        satConstellation:part:2
        satConstellation:part:3
        satConstellation:part:x(<=9)
Hakansv commented 3 years ago

The gnss type and satellites used for fix calculation is present in xxGSA. So I intend to add also that sentence. Questions:

type:   
   meta:    
      description:       "Fix type"
         value:          "3D"
         $source:        "blueconv.GP"
         timestamp:      "2021-01-09T15:18:30.153Z"
         sentence:       "GSA"
usedIDforFix:   
   meta:                 {}
         value:          "24, 17, 19, 12, 15, 25, 32"
         $source:        "blueconv.GP"
         timestamp:      "2021-01-09T15:18:30.153Z"
         sentence:       "GSA"
Hakansv commented 3 years ago

Many Qs now... I found a meta description file here: /usr/lib/node_modules/signalk-server/node_modules/@signalk/signalk-schema/dist/keyswithmetadata.json Is that the correct file? I saw that description was also included in files for "vessels" and not "self" as the above. If so it's not included in the NMEA0183>SK project but has to PR to SK itself. Or?

For a test I used this meta descr. Would that fit? (The path itself would possible be changed I suppose.) "usedIDforFix":{"meta":{"description":"Satellite IDs used for Fix calculation"},

Hakansv commented 3 years ago

re: "Note the talker ID to distinguish between the systems:!" some posts above Neither the NMEA0183 4.11 seems to message the gnss system. See this But, GPS satellites has satellite ID from 1 - 32. Higher IDs are from the other system so that will be the way to distinguish for now. See the above linked doc for details.

tkurki commented 3 years ago

I don't think separate constellation elements 1-3 are a good idea. The division to several sentences/lines is just in 0183 data. In NMEA2000 it is a one or more pgn 129540 messages. We just need to parse several GSV lines into a single delta, just like AIS messages are parsed.

Furthermore 129540 seems to have also data about the use status of each satellite. I need to dig a little deeper and check also other data sets. One possibility is that in nmea0183-to-signalk parser we combine GSV and GSA sentences to a single delta, that is pretty close to PGN129540.

Thanks for the link to NMEA0138 Errata. I did not understand

Neither the NMEA0183 4.11 seems to message the gnss system

because the document repeats what I wrote above about talker id being significant.

That document also says To reliably identify when the satellites in view were observed a corresponding transmission of the ZDA sentence is recommended.. So in effect the GNSS is saying "at time xxx I saw satellites a, b and c from system A and x, y and z from system Y and I used a and b for the fix.", but that is split to several sentences in 0183.

Mind if I take over from here? We need to make the parser stateful, so that it can combine several sentences.

Do you happen to have any data with tag blocks that work in the way described in the errata? Sounds like tag blocks are used to tie together the sentences into a single unit.

tkurki commented 3 years ago

keyswithmetadata are generated from the specification schema files - imho we should first get this working in 0183 and n2k and only then add this to the schema.

tkurki commented 3 years ago

Oh right, the one versus many PGNs is just an artefact of how the inputs that I am looking at are generated.

Hakansv commented 3 years ago
Hakansv commented 3 years ago

re: " One possibility is that in nmea0183-to-signalk parser we combine GSV and GSA sentences to a single delta" But don't rely on GSA is sent. I'd say it's seldom used. It depends on the GNSS device. It's not present from my onboard GPS.

tkurki commented 3 years ago

tag block is a mechanism to put additional fields in the beginning of NMEA0183 data line, see https://github.com/SignalK/nmea0183-signalk/blob/master/test/tagblock.js#L24

You need to reread the bulletin, there are three parts

tkurki commented 3 years ago

Yep, may need to add an option to combine the sentences.

Hakansv commented 3 years ago

I'll read and learn and look forward to study your solutions. About PGN I've no particular experience from N2k yet. Not more than some test with Openplotter and SK. If needed I'll test when you need and also consider a OCPN parsing of the gnss data from SK-GSV. That was one of my goals with this. Thanks Håkan

Hakansv commented 3 years ago

Just for info I think a single delta for all (<=12?) satellites will be manageable for OCPN Dashboard. I just have to group them up to four each, as the GSV does. The same instrument coding will be used both for SK and GSV data. But no problem.

larsth commented 3 years ago

Regarding help wanted to create a schema for satellites in view, A program called gpsd has JSON output that can easily be transpiled into something else, i.e. a SignalK message

You can look at my repositories and find them using a browser search using "gpsd" (without the "s of course) My repositories : https://github.com/larsth?tab=repositories

FIY, I had jumped into this conversation without reading it, becuase SignalK looks interesting.

Edit 1: Jumped, not jumping