Reported SNR of GPS Receivers

On the ones I've designed there was an RSSI function (in the 2nd IF). I don't know if the software developed an SNR "report" from that or not. ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |

      Remember: Once you go over the hill, you pick up speed

d

Higher end (including aviation) GPS have SNR reporting. It is important to monitor the signal to noise ratio degradation on airplanes when other avionics are turned on.

I believe the SNR calculations are done by looking at the stability of the data at the very end of the receiver chain.

Given the nature of the signal, I kind of figured that would be the only useful SNR estimate.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" was written for you.
See details at http://www.wescottdesign.com/actfes/actfes.html

...

I would *guess* that the variance of the reported position

*formally* depends on the SNR + system-specific factors. If one were able to control the system-specific factors 'sufficiently well' one would probably be able to estimate SNR from variance in position.

I have no idea what it would takes to do this - probably some sort of calibration in a controlled environment.

Rune

d

The SNR appears for each satellite on most GPSs, well at least what I've owned in the last decade. There are algorithms to weight the information of each bird based on SNR before determining position. The timing GPS can do this to, but most use an elevation filtering scheme rather than SNR. They ignore birds on the horizon and up to a cutoff the user selects.

There is a usenet forum for GPS.

sci.geo.satellite-nav

There is very good or very bad info there.

Isn't the data in a digital format before it is used to determine position? I don't see how you can determine signal to noise based on errors in a digital stream - rather, you can determine error rate, but that is distinct from SNR.

Also, doesn't position depend upon the combination of several inputs? It doesn't make sense to use variance in position to determine SNR when you can't really know which of the many inputs are responsible for the error.

Brian Willoughby Sound Consulting

of

The error rates depend on the SNR of the recieved signal. If the error correction system is seen to be too busy correcting errors, it's at least an indication that the numbers in the associated data stream might be viewed with some suspicion. An error correcting system can only handle so much - there might be errors coming through if the SNR is low enough.

Rune

Usually you'd take the decoded data pattern, compute the expected constellation points, and compare those to the received signal at the demodulator.

I don't recall the GPS spec having an SNR measurement requirement. Some other standards have such a beast.

Steve

I don't know enough about the specifics of the GPS signals to say for certain, but usually there are multiple ways to estimate SNR depending on the signal and what you want to do.

The received constellation can be sliced so that the error vector is measured from the sliced constellation point. This works well from mid- to high-snr but when the noise starts saturating in the dynamic range of the slicer the quality of the estimate degrades pretty quickly.

There are a number of techniques where the incoming raw error rate can be estimated and then compared against the matched filter bound for that modulation type. That's often reasonably accurate across a very useful range of SNR.

For some signalling, and I think GPS falls into this category, there are chunks of the signal that are known apriori, either Unique Words for framing, or chunks of known sequence used for synchronization or channel estimation, etc. Since the sequences or UWs are known and appear in known locations in the signal, they can be used pretty effectively for SNR estimation if the noise statistics are stationary enough (which they usually are, and I'd expect so for GPS).

Eric Jacobsen

If you're doing it at the receiver you should be able to do things like compute the variance of the amplitude of the output of the integrate-and-dump right before dump, or like Eric said, take advantage of known data sequences.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" was written for you.
See details at http://www.wescottdesign.com/actfes/actfes.html

That's all of Usenet in a nutshell.

d

Here is a little more info:

In GPS world the "SNR" is discussed in terms of C/N or carrier to noise level. Look up carrier to noise on google and you will get a little help.

The book : Undersatanding GPS: Principles and Applications by Elliot Kaplan (he is the credited author, but it is compilation of many authors works) Is the definitive book on GPS when diving into the engineering aspects of how to design GPS receivers.

I worked on some of this stuff 5 years ago or so, but my memory fails me on the details.

Or the whole internet! ;)

Eric Jacobsen

When the GPS system was first revealed to the general public, there was claims of systems working down to -20 dB SNR :-).

Of course, this was calculated comparing the signal power to the thermal noise in 1.023 MHz bandwidth. However, since the actual despread net data rate is only 50 bit/s, that -20 dB SNR would translate to something like +23 SNR :-).

I designed a PLL data extractor that could function on a TACAN signal at around -6dB SNR (all analog, of course ;-). That's the one that Gardner (of PLL book fame) opined couldn't work, until I demonstrated it to him.

I was 28 at the time, and a wee bit scared of confronting Gardner.

But I had already cleaned Larry Huelsman's clock a few years before, proving his paper on desensitizing NIC's was a bunch of crap, so I was experienced :-) ...Jim Thompson

--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |

      Remember: Once you go over the hill, you pick up speed

That because it is from a satellite, and in the satellite world, everything is noise, even the signal.

A valid term here is "noise floor". Then, there becomes a point at which it invades your "good noise" so much. Then, "bit error rate" becomes a valid term. That threshold is more important than s/n numbers.

The carrier is analog, but all of the signals are digital, heavy FEC laden packets.

I am currently working on our battlefield tracking and comms solutions.

Our GPS antenna is a very precisely positioned octal array. Of course the whole system goes far deeper than the antenna.

But ESPECIALLY Usenet, and usually toward the latter type.

That being the case, how much can you count on my opinion... either?

The way is that *some* of us can see the glimmers through the sewage waterfall.