Here's What's Up with SVN49

November 23, 2009 By: Alan Cameron

GNSS Design & Test Newsletter, November 2009

Last week I conducted a technical webinar on SVN49, and this column consists of the audience questions and panelists’ (Peter Boulton from Spirent and Greg Turetzky from SiRF/CSR) answers to them. You can download the audio potion of the live webinar and all the accompanying powerpoint slides from the first news story in this newsletter, below. It is very comprehensive and should be helpful to anyone involved in system design or receiver design. I found the Q&A session particularly stimulating, so I’m putting it online in written form here. We start out with “Why are some receivers not affected by the SVN49 signal anomaly, and others are?”

Before moving to the questions, here’s a brief digest of what we covered in the first 40 minutes, which, again, you can download at www.gpsworld.com/webinar.

My intro:
SVN 49 Brief Background
SVN 49 Problem
SVN 49 Cause
SVN 49 Proposed Remedies

Peter Boulton’s slides:
Call for Help
Simulating the Anomaly
Create Scenario
Generate Signal and Reflection
Forcing On-satellite Reflection and LOS
Scenario View
Reflected to Direct Power
Simulating the Variation with Elevation Angle
UCD file structure –Varying the power level
UCD file structure –Phase Inversion
Scenario running
Removing the Anomaly from Simulation
Partial Fix?
Simulating the Phase Centre Offset
Simulating the Undeclared Clock Bias
Summary

Greg Turetzky’s slides:
What are important attributes of consumers?
Important attributes of consumer chipsets
The “partial fix”
Going forward

Now the Q&A, verbatim.

Alan Cameron (moderator): Our first question, in fact, leads into something I wanted to ask. “Why are some receivers not affected by the problem, and others are?”

We’ve just heard from Greg Turetzky of SiRF Technology/CSR plc, that the type of consumer-oriented receivers that they manufacture are largely not affected, or not noticeably affected by the SVN49 signal anomaly: they don’t see the multipath element. On the other hand, high-precision receivers in other market segments are very much affected. The third type of receiver, the one we don’t know about, but which will probably be the key determining factor in what is done with SVN49, what is done with its signal: How are military receivers affected?

We can only presume that the Air Force and other arms of the U.S. military are doing some testing now to figure that out. That will certainly be a key indicator in the final answer to this question.

Before I turn this question over to Peter and Greg to answer, I’d like to show an example of capability to cope with effects of some advanced receivers, some high-precision receivers. Here we see a plot of the SVN49 signal showing the error.

Now some receivers can see the error, and some receivers with their advanced multipath mitigation can strip out the error. We have seen demonstrated results using JAVAD GNSS Triumph receivers, as shown here.

At the special panel session on SVN49 at ION in September, Javad Ashjaee stood up and proposed his solution to the dilemma: “Just turn it on. Let’s go. Let’s use this as an opportunity to study multipath.”

Well, he’s in the position of having and marketing receivers that can cope with it, but this is not true across the board. It may be true of some advanced high-precision receivers, it may not be true of all of them. Given that kind of broad range of the different kinds of receivers in the field, gentlemen, let’s go back to the original question: Why are some receivers not affected by the problem, and others are? And then how does this influence things, in your mind, as to how this might develop going forward?

Greg Turetzky, SiRF/CSR: There are two things that impact it. One is what I talked about, which is that the bandwidth of the receivers is different, and therefore the amount of multipath error that you see is different, depending on the bandwidth of the front end. The other thing is what Javad demonstrated, which is the correlator spacing and the ability to mitigate multipath at the receiver level is another way. Similarly, like Javad said, we’re all going to discover great ways to mitigate this problem going forward, at the receiver level. Some of those mitigations already exist, and I’m sure people will discover more. It basically has to do with receiver design and the fact that we have this extra bandwidth and we have different ways of using the correlators to determine the shape of the incoming correlation curve, and therefore those receivers will react differently in terms of their ability to see it, and also to compensate for it.

Peter Boulton, Spirent Communications: It’s largely a problem of the correlator spacing, as Greg has said. Sometimes it will fall within the correlation space, and sometimes it won’t. You’ve got to remember this is a fixed multipath, with a 30-nanosecond delay. It’s always 30 nanoseconds, so if your correlator is narrower than that, you’re just not going to see it. But if you use a traditional early-late correlator in a very low-priced receiver, then it’s just going to fall within the correlator and it’s going to bias the correlation peak to one side, and that’s why you see this error of four meters or so.

I can also tackle another question that I see submitted here, and that is: "Why is the error elevation-dependent?"

This has to do with the fact that the transmitter onboard the satellite that is broadcasting this multipath is a narrow beam, outer ring of the antennas, which has a much narrower beam, and at lower elevations you just can’t see it. You’re looking at the side lobes of the transmissions, which means that the amplitude is very much lower, and it doesn’t impact that correlator. So although the signal appears in there, its amplitude is so low, you can’t see it. But as the elevation angle increases, you start to see the signal through the main lobe, until at zenith you see the full impact of that signal at L1, 13db down .

Alan Cameron: I see a question here that I can answer, because there is no answer. Someone has asked: "How soon will the GPS Wing set SVN49 healthy?"

The answer that we have received to that question, it was presented in Stockholm when I was there for the International Association of Institutes of Navigation World Congress in late October, the answer is not definite, but what the GPS Wing and other U.S. government PNT representatives stated was that “it’s not a question of if, but when.” They have ruled out, apparently, discarding the satellite; they will set it healthy, and indications are, though this was not strictly speaking officially stated, sometime in 2010. It’s just impossible to predict how long it will take them to complete testing on all the military receivers on different platforms, and compile manufacturer input and so on. That question has no real answer, as yet, at least.

I want to turn to another question, which kind of hits at this. "If using only a P-code receiver, can the effect be differentially corrected?"

Peter Boulton: Once again, if the signal, the multipath signal, falls within the correlator spacing of the receiver, it will impact the correlation peak and give you an error. Now if you noted earlier in the presentation, I showed that the L1 and L2 components of the reflection were at different levels, and I also showed that the beam width was different for L1 and L2. That means that the impact is different for L1 and L2. So if you had a P-code receiver with L1 and L2 and you were using it differentially, what you’ve now got is the differential error. If it was common at both, you might be able to back the error out somewhat, by looking at your L1 and L2 measurements. But in fact, this introduces further error on differential receivers that have this signal appearing within their correlators. So no, just P-code alone doesn’t get you a fix, at all. What you need to do is eliminate the signal from the correlator space.

Alan Cameron: Another listener question, and we may have addressed this in passing: "Does the problem also exist in carrier phase of the signal?"

Peter Boulton: That’s in the same vein as the question about why some receivers are affected more than others, and what is the cause of this observation. It’s because it’s falling in the correlator and offsetting the peak of that correlator, introducing a pseudorange error. And that is the biggest problem with the signal, that’s how it creates the impacts we’re seeing. In carrier phase, the impact is much lower, unless you start looking at differential carrier phase. And because of the anomalies between L1 and L2, then you might be introducing errors by looking at the carrier phase. But if you were an L1-only observer looking at carrier phase, it really doesn’t have a great deal of impact. It’s in the pseudorange, the code phase, that you’re going to see the biggest problem.

Alan Cameron: Regarding the Spirent scenarios you briefed us on earlier in the webinar, Peter. One listener asks: "What do the scenarios consist of? Simulated pseudorange and carrier phase measurements for a ground station?"

Peter Boulton: They are a static observer. As I mentioned in the presentation, we found a spot on the Earth for our nominal simulation where the satellite was directly overhead, and we’re using the motion of the satellite to introduce the variability throughout the simulation. So we’ve got a static user at that location, the satellite is immediately overhead, and then it just descends to the horizon over a period of about 3.25 hours.

Alan Cameron: I’m not sure if either of you can address this, because it has to do with the satellite operation itself. "If the satellite has some sway motion, how will the lever arm difference between the antennas affect the signal?"

Greg Turetzky: Let’s put back up the slide that showed the inner ring and outer ring antennas.

I would say, from a first pass of looking at it, those distances aren’t sufficiently large to have an impact, because those scales are labeled. But not being an expert in satellite dynamics, I wouldn’t pretend to answer it except to say, I think this is the chart that gives you the best clue, that I would expect it to be small.

Peter Boulton: Effectively, the satellite sway is equivalent to you having a change of elevation angles, because effectively you’re looking at the transmission patterns of that satellite, and it would have to move quite a long way for you to move down that elevation curve, and I don’t suspect that the satellite moves that much. So I would agree that the impact of a sway would be very small, particularly when you’re close to the bore sight. So if you’re above 60 degrees of elevation angle, you’d see virtually no impact. Now the signal would change much more significantly closer to those nulls at the pattern at 39 and 41 degrees, so by then the amplitude of that reflection is small anyway, so again you wouldn’t see much impact due to the motion of the satellite.

Alan Cameron: One final question again pertains to satellite design: "How did you calculate the delay of 29.82ns from the 2x162 inch cable? Which velocity factor (or permittivity) of the cable within the satellite have you used for that calculation?"

Peter Boulton: We didn’t do anything particularly scientific at that level. We had the two statements in the USAF data that there was a 162-inch cable involved, and that the delay was approximately 30 nanoseconds. So here’s the chart again that I used to present that.

We had a choice here, because 8.9384 meters is not 2 x 162 inches. If I actually use the 162-inch value and then put the reflected signal in phase, I get a delay which is shorter than 29 nanoseconds, considerably shorter, several nanoseconds shorter. I took the decision to use the 30-nanosecond delay as a guide, and just calculated how many whole wavelengths at L1 that made up, and worked out that it was 8.9384, which would put the signal in phase. The 162 inches was just a guide that was in the data, and I focused on the 30-nanosecond number. I didn’t go anywhere into satellites and that sort of thing - cable permittivity. It’s just looking at that observation, it’s the result of that signal reflection through that cable and is related to its length. Its permittivity is related to the observed power difference between the line of sight and the reflected signal, which is varying with elevation angle. It’s largely irrelevant to the simulation side of things.