Network-RTK (Ideas/Hardware/Software)

Hello everyone,
I want to build up my own Rtk Base.
I like the Idea to be independant of any correction Service.
So I have read through several documents.
There is mentioned with an Single Rtk base its just possible to use it with in 5 to 20 km.
So my thought is maybe that some Farmers built up a own Rtk Net. The minimum need is 3 RTK Base Station. With an Distance of max. 50 km.
Within the Triangel the Korrection due to Atmosphäre failures will be interpolated.
I am pretty New to this topic, maybe some likes the idea and we can go on.
Passion works! Lets see what we can do.
Regards, Peter

p. s. Some links to read through.

https://www.google.com/url?sa=t&source=web&rct=j&url=https://leica-geosystems.com/-/media/files/leicageosystems/products/manuals%2520quickguides/guide_to_reference_stations.ashx%3Fla%3Den%26hash%3D87D34643C5113316755664C809EEC74B&ved=2ahUKEwjqrrSAneTvAhWKPewKHRotC5IQFjACegQIGBAC&usg=AOvVaw2rQPj_79xF5FQIQ_iteuDF&cshid=1617528024864

https://www.intechopen.com/books/satellite-positioning-methods-models-and-applications/network-real-time-kinematic-nrtk-positioning-description-architectures-and-performances

For the Ublox F9P, it will hold a baseline for 100km. The stated accuracy of 10mm+1mm/km for the first 20km is stated in their manual, but will hold a fix far outside this.

If you do not need the extreme sub cm accuracy, using only one cellular base could be a very easy option.

For pass to pass accuracy you also do not need absolute accuracy of your base coordinates, whatever the base surveys in is fine as long as you write it down and use the same base coordinates for the field you are working in year after year. If you are not using tramlines or trying to run the same sprayer tracks and start fresh every time, just surveying in new at location every time is fine.

Lora radio, or other data radio types are also very useful. But usually require line of sight to work, which will involve setting up a tower at the base. This is my favourite option, no bill, fairly long distance, ability to use custom refresh rates.

Alternately you can setup Base Monuments at different fields and bring the base with you connected locally through radio.

For high performance RTK applications nothing is better than radio, Transmission is almost instantaneous, you are not limited by data charges for high correction refresh rates and using all GNSS constellations.

Minimum correction data is fine 1Hz for each constellation used, 0.1hz for ARP coordinates for 99% of farm work. High refresh corrections at 10Hz arp still at 0.1hz, are only really beneficial for high accuracy demands like dozer blade control, and laying tile. But they are fun to play with.

According to F9P manual corrections should ideally be sent to the rover within 50ms. So having to make a hike to an NTRIP server on the internet and back usually has a penalty of 60-300ms. NTRIP is a very convenient way to handle corrections but is slow. Will it affect most farming operations running stale corrections? certainly not, so ntrip is fine.

The biggest question to ask when setting up a new base is, “What is the maximum amount of error i am willing to put up with?”

Buying commercial corrections for RTK, in Alberta for example, you may get stuck using a base station in the nearest town over +50km away or more, with no VRS. Trimble equipment currently can hold an infinite baseline, but not a very accurate one at extreme distance. But its still full price, lol.

1 Like

You may try back40precision software which can be downloaded at https://www.back40precision.com/. It offers Nearest, Single base, iMac, Mac and FKP solutions. The inputs from the bases are either RTCM 3.0 or RTCM 3.2 MSM4 on TCP or via NTRIP server.

The only drawback is that it doesn’t work with RTCM 3.2 MSM7.

1 Like

Could RTKLIB be used to translate MSM7 to MSM4?

A quick look at the caster manual gives the impression that they support only GPS + Glonass, do they actually support Galileo and Beidou too?

Yes, STRSVR module does it nicely. A nice workaround though is to configure the F9P to output MSM4.

It does support all, GPS, GLONASS, BDS and Galileo. The only problem I have noticed is that the Rtcm3. 0 output in single base or nearest just outputs GLONASS observables only! But RTCM 3.2 is OK.

1 Like

I was more thinking about using Euref streams. Many of them are MSM7.

Try STRSVR to downgrade the streams. The algorithm works pretty first. No noticeable delay.

You can also set strsvr to output the corrections at a lower frequency/rate then the standard 1hz, for whatever reason you can think of.

I guess you can’t use MAC/MAX/FKP corrections with the F9P? VRS is easier as the receiver doesn’t know if it’s a virtual or real station.

Can you translate that to plain English. :slightly_smiling_face: I did not see VRS on the list on the earlier post. My interpretation is that “single” or “nearest” are not VRS but just picking up one of the available base stations. Do you mean back40precision caster would not help with an F9P if one wants to build a better (VRS type) correction signal from multiple base stations that all are a bit too far away?

Here is a response from Geo++ when last week I enquired then about their network RTK software cost.

"The price for GNSMART software license is individually calculated and based on:

the number of reference stations,
the size of the service area (measured in square kilometers),
the type of service: either OSR (observation space representation, classic RTK corrections) or SSR (state space representation, e.g. PPP-RTK), or both OSR+SSR.

Could you please send us this information? We will send you an official quotation as soon as we receive your answer

My collegue Andre Warneke will answer your technical questions

Best regards

Isabel Frost"

I will respond there now with 3 base stations and some 500 square kilometre coverage (to fill in the worst gap in the middle of the base station triangle).

In a Turkish test they found that up to 40 km distance single base was actually better than network (which had larger error at short range). At 60 km distance network started to be a bit better than single base.

I will get back here after I hear from them. That is the professional solution, but I fear it is a bit pricey :grimacing:

1 Like

Maybe it would be easier calculation vice to make one virtual base station in the middle of an existing coverage gap, instead of customised virtual base for each user.

3 Likes

I’ve been trying to find some good research papers on the algorithms, with little luck

Some are as easy as interpolation, but before that some magic needs to happen regarding the ambiguities between the RTK stations, not well enough educated on RTK things to tell.

As I read the manual it only supports nearest or max/mac/fkp. These three are network RTK but work different from VRS, they send more correction stuff to the receiver and the receiver does the work, whereas in VRS the station does the work for the rover.

1 Like

So… I got a quotation from Geo++ for a network RTK calculation software lisence (GNSMART). Modest one time fee of 50 k€ + 7k€ per year for updates & follow up or 12 k€ rental price for one year at a time :person_facepalming:

Government agencies are being ripped off as usual :woozy_face:

On the other hand, the recommended hardware is minimum 8 core CPU, 12 MB (L3) cache etc means there is some rather heavy number crunching involved.

1 Like

can you get into more details ? Should the simple param -msg "1074(1),1084(1),1094(1)" work for downgrading MSM7 to MSM4 for str2str ?