Upsampling

[Steve Chatelle]

We tried the upsampling with the LS as a network rover and were shocked at the upgrade in performance. We have a free and powerful RTN in New York, but who operates at 1 Hz anymore. Having this 5 Hz option at hand as we reach the ends of our radio range will change our methods for sure.

Bigger for us though is the upsampling of our UHF. We often run two or three rovers from a base, but at the 5 Hz that the LS demands now, our 2008 vintage Triumph-1 rovers are usable but pretty flaky. In the conditions where the Tri-1s would have stay fixed for hours, at 5 Hz broadcast they go autonomous as soon as the rover moves. Once we stop, they slowly cycle autonomous to fixed with 7 or 8 sats to fixed with 10+ sats. The positions stored then are almost always good.

This morning, though, I copied our 5Hz profile to a 1-5Hz profile, only toggling on the upsample. I set up a Triumph 1 base using SurvCE just to test, broadcasting at 1 Hz of course. Walked over to the LS and it was already fixed with 6 engines. Stayed fixed against the building and fixed quickly under the trees at 5 Hz upsampled. A Triumph1 rover also fixed and stayed fixed the whole time, like old times.

So, a question we have is theoretically how much performance are we sacrificing by upsampling 1 Hz vs. broadcasting 5Hz?

You know, just as you get confident that you have mastered and implemented the state of the art of RTK, Javad quietly re-states the art. I couldn't be more happy with this equipment.

 

[Nate The Surveyor]

Rumor has it that Javad is re-inventing the wheel again... (Like that was news!)

Seems that he is going back to 1 hz.... BUT this is only at the base.
This will give us our 1 hz range back....
and, longer battery life, at the base.

HOWEVER,
at the rover, he has it going at 5 hz still... so this will somehow be based on (I'm guessing here) a weighted interpolation of the data, from the 1 hz data. So....
This could mean that if you are running a network rover, that you will get the 5 hz performance! (I hope this is true). I don't have a network here. But, for those of you who have it... this would be incredible!

Sometimes, when I am in the DENSE stuff, my radio says it is 1 hz, then 5 hz, then 3.5 hz. I think it is due to the enormous amount of data, that 5 hz at the base gives us.... and when it gets bouncing around in the woods... deteriorates pretty fast.
I hope this fixes that problem.
I hope that we get our long range back.
I am VERY excited about this new development.

I want to take advantage of EVERY technological advance... I spend the 1st 1/2 of my life, with outdated equipment... I never want to practice that way again.

Nate

 

[Matt Johnson]

So, a question we have is theoretically how much performance are we sacrificing by upsampling 1 Hz vs. broadcasting 5Hz?

There is no performance sacrifice. I confirmed this by using an external GNSS antenna split to two LS's. Both units were receiving corrections through TCP from the same T2 base station but one LS was receiving 5 Hz corrections and the other was receiving 1 Hz corrections that were then upsampled to 5 Hz.

 

[Steve Hankins]

There is no performance sacrifice. I confirmed this by using an external GNSS antenna split to two LS's. Both units were receiving corrections through TCP from the same T2 base station but one LS was receiving 5 Hz corrections and the other was receiving 1 Hz corrections that were then upsampled to 5 Hz.

Is the upsampling filtered or unfiltered? I am trying to understand how we get more from less. As always, thank you for what you do, great equipment.

 

[Matt Johnson]

Is the upsampling filtered or unfiltered? I am trying to understand how we get more from less. As always, thank you for what you do, great equipment.

I am not sure I understand what you mean by filtered or unfiltered.

 

[Steve Hankins]

I am not sure I understand what you mean by filtered or unfiltered.

Matt, thank you for the response. After further research I am not sure upsampling filters would apply to this situation. Maybe the better question for me to ask is the upsampling a straight linear interpolation between the 1Hz. data packets broadcast from, in my case, the base and uhf radio, to 5 Hz. data at the rover?

 

[Nate The Surveyor]

This whole deal of "Upsampling" appears to be a drastic departure from the previous goals... ie, actual use of 5 hz at the base, and rover.
It was conveyed to us (the users) that there was a major difference between "Interploated 5 hz" and actual, from the base.
Now, we are curious.
Thanks.
N

 

[Shawn Billings]

Actually we warned against extrapolated corrections.

 

[Steve Hankins]

Actually we warned against extrapolated corrections.

Shawn, what are your thoughts on Interpolated corrections as opposed to Extrapolated corrections with respect to the actual 5Hz. broadcast?

 

[Jim Frame]

I'm still trying to figure out how to configure Verify to take more than 10 epochs. Confidence Level doesn't seem to be that setting. Does it reside on some other page?

 

[Adam]

I'm still trying to figure out how to configure Verify to take more than 10 epochs. Confidence Level doesn't seem to be that setting. Does it reside on some other page?

Jim, I can go over the settings with you if you can get on Rams.

 

[Nate The Surveyor]

So, interpolation is "Data BETWEEN real data points". And, Extrapolation, is "data PROJECTED forward of real data points". So, in theory, use of Interpolated data, requires a little more delay, than extrapolated data. However it can be more accurate, because it is based on data on BOTH sides, and Extrapolated is merely a PROJECTED data point, based on real data points.

I really don't care... about HOW.... I care about WHAT. And, what I want is GOOD data FAST.

 

[Matt Johnson]

This whole deal of "Upsampling" appears to be a drastic departure from the previous goals... ie, actual use of 5 hz at the base, and rover.
It was conveyed to us (the users) that there was a major difference between "Interploated 5 hz" and actual, from the base.
Now, we are curious.
Thanks.
N

The major difference is that we are attempting to fix the ambiguities at a 5 Hz rate so a fixed solution can be obtained much quicker. Most RTK systems with update rates greater than 1 Hz still fix the ambiguities at a 1 Hz rate but then once a fixed solution is obtained, the data is extrapolated to get positioning rates of up to 100 Hz. Upsampling 1 Hz base data allows the ambiguities fixing process to occur at a 5 Hz rate.

So, interpolation is "Data BETWEEN real data points". And, Extrapolation, is "data PROJECTED forward of real data points". So, in theory, use of Interpolated data, requires a little more delay, than extrapolated data. However it can be more accurate, because it is based on data on BOTH sides, and Extrapolated is merely a PROJECTED data point, based on real data points.

Yes this is correct.

 

[Matt Johnson]

Maybe the better question for me to ask is the upsampling a straight linear interpolation between the 1Hz. data packets broadcast from, in my case, the base and uhf radio, to 5 Hz. data at the rover?

The Upsampling process is a proprietary process that was recently developed so it can't be disclosed exactly how it works, you just need to know that it works and the ambiguities are being resolved at 5 Hz with 1 Hz corrections when it is enabled.

 

[Steve Hankins]

Roger that, and thank you for the reply. In the case of changing the broadcast from the UHF radio, ( 1 watt), back to 1Hz., what would you recommend for bandwidth and modulation using the upsampling?

 

[Shawn Billings]

Bandwidth must be 12.5khz (narrow band). I would recommend DQPSK for modulation, but I have one customer who appears to be using DBPSK with success, so you might give it a try.

 

[Matt Johnson]

In the update that will be released, the modulation choices have been limited to D8PSK and D16QAM so D8PSK would be used for 1 Hz corrections:

 

[Steve Hankins]

Thank you both.

 

[Jim Campi]

This is apparently another significant benefit of upsampling. As you probably recall, one of the only drawbacks to broadcasting at 5 Hz is decreased radio range/penetration. To achieve the bandwidth necessary to transmit the corrections from our base at 5 Hz we were required to modify the modulation to D16QAM.

We now have the option of transmitting from our base at 1 Hz, upsampling at the rover and running the corrections through the Beast mode processor at 5 Hz with no loss in fidelity. Upsampling effectively neutralizes the only drawback in processing at 5 Hz.

By running the base at 1 Hz with a modulation of D8PSK, upsampling at the rover and processing at 5 Hz, we regain the loss in UHF radio range. I use the 1 watt radio so heat has not been a problem. For those of you transmitting 35 watts at 5 Hz, you may have experienced additional heat management issues. I suspect the battery run time may increase as well.

Has anyone tried this configuration?

 

[Nate The Surveyor]

I am running 35w radio. Today I will report back. I have the fan. I suspect I don't need the fan....