AIS ZIG-ZAG

[cagney]

We were anchored off Isle of Wight a few days waiting for better weather.
Notice the zig-zag in the Wight Lines, Portsmouth to Wootton Crek ferry, "St Faith"? Notice also how the pattern seems to be different at high and low speed. Time dependent!
I have never seen that before.A few other vessels, including a pilot boat displayed this pattern.
Cold it be a software problem with the vessels AIS-units? But, if so, why haven't I seen it before... or since.
Radio interference on-board the transmitting vessels, perhaps?

Thomas

[bcn]

Hi Thomas,

does your own GPS jump in this zone? The spikes are not that big (15m?), aren't they? Spikes would explain why the amplitude is more or less the same and the frequency (of the graph) goes down with higher speed.

Would be funny to see if the spikes are even distributed over time (Lighthouse pattern)

Nice catch...

Hubert

btw.: COMAR, the AIS equipment manufacturer is based on the Isle of Wright. Perhaps they know what is going on on their backyard...

[MarkJ]

I have seen it a few times, not often, but enough to know its some aberration from the transmitting vessel.

[bcn]

The issue here is that Thomas did observe this signals from various vessels.
Or all have the same sensor with the same problem or... ?

[svlamorocha]

Originally Posted by cagney

We were anchored off Isle of Wight a few days waiting for better weather.
Notice the zig-zag in the Wight Lines, Portsmouth to Wootton Crek ferry, "St Faith"? Notice also how the pattern seems to be different at high and low speed. Time dependent!
I have never seen that before.A few other vessels, including a pilot boat displayed this pattern.
Cold it be a software problem with the vessels AIS-units? But, if so, why haven't I seen it before... or since.
Radio interference on-board the transmitting vessels, perhaps?

Thomas

I doubt RFI will cause an error like that given that AIS position is transmitted over the air not as analog data but as digital data modulated using Gaussian minimum shift keying. If there is an error induced by RFI that still goes across as fake valid data the odds are that the point would be anywhere else but close to the correct trajectory.

I bet there is a glitch in the OpenCPN algorithm that draws a smooth line based on discrete data points received over AIS. I have written some lousy smoothing algorithms myself in my youth. If you gran teh NMEA sentences showing the successive positions of that boat and plot them you will probably find out the answer-

[cagney]

My own GPS receiver was steady in this area.

Study the ferry track. There is a marked change in track pattern around the N cardinal buoy.
The ferry reverses out of Wootton creek and turns in the vicinity of this beacon. So .... only "zig-zags" when the ferry moves forward.


Thomas

[PjotrC]

I have such a case recorded, so I could look at it with a couple of plotters and tools... I am pretty confident the zig-zag occurs at the transmitting side. Of course, it much less visible at higher speed, but also does happen. In my case the most serious zig-zagging also started when the vessel abruptly stopped and resumed speed in another direction. Typical AIS position accuracy (Class A, Type 1 positioning, GPS, No differential) is officially "not better than 10m".

[piluso]

Looks as GPS heading error at the AIS transmission side, similar to goto mode in a handheld GPS, where the needle goes around the heading. Probably a faster speed gives to GPS an improved heading calculation; so the error angle of each "leg" decreases with the speed.

[PjotrC]

Originally Posted by piluso

similar to goto mode in a handheld GPS, where the needle goes around the heading

Yes, it does look similar, but in my case all other parameters (COG, HDG, SOG,...) are transmitted without such "jumping" effect. Indeed one would have to look into the transponder code...

[svlamorocha]

Everything I here seems consistent with a plotting algorithm that tries to fit a track line that gives too much weight to the jumpy instantaneous COG course data that is sampled by the AIS transceiver, relative to the weight given to the position data. I bet the wiggles almost go away if you just joint the position points with a polygonal.

If there is a volunteer who can capture the AIS NMEA data for one of these ferries from a receiver I will be happy to do the investigation.

Sent from my GT-I9192 using Cruisers Sailing Forum mobile app

[PjotrC]

Originally Posted by svlamorocha

I bet the wiggles almost go away if you just joint the position points with a polygonal.

In my example, they don't go away ...

It is true that OCPN-generated track is not exactly the same as the polygon linking all positions, but the difference - comparing to hand-generated exact polygon - is not really smoothing, and the wiggles still remain. I checked this also in NavMonPC and NavSim, and they all produce wiggles.

On the transmitter side, I am not quite clear about what the reason for zigzagging is. HDG is taken from an external gyro sensor, so no spikes there, OK. But SOG and COG is computed inside from lat-lon. Obviously SOG and COG is smoothed somehow before transmitting, otherwise it would have to jump as well.

Two reasons for lat-lon inaccuracies are 1) inherent GPS inaccuracy 2) sideways movement of GPS antenna mounted very high above water level on a big ship (especially when turning at low speed). Together these two might possibly account for being 5-8 meters off every time... but it is strange to see the port-stbd deviation regularity of successive position reports. I would expect it to be more random.

Ps. In some applications it might be useful to have OCPN produce tracks really identical to successive positions reported. (1.3.6 ?) ...

[bdbcat]

Thomas, et al...

Good to hear from you. Hope you are having a good cruise, without too many boat problems.

Re: zigzag...

OCPN does no filtering of data coming from remote AIS transponder. We simply capture the lat/lon as it comes it, record the points in a list, and render it like so:

       if( (!b_noshow && td->b_show_track) || b_forceshow ) {
            wxPoint TrackPointA;
            wxPoint TrackPointB;

            dc.SetPen( wxPen( GetGlobalColor( _T ( "CHMGD" ) ), 2 ) );

            //    First point
            wxAISTargetTrackListNode *node = td->m_ptrack->GetFirst();
            if( node ) {
                AISTargetTrackPoint *ptrack_point = node->GetData();
                cc1->GetCanvasPointPix( ptrack_point->m_lat, ptrack_point->m_lon, &TrackPointA );
                node = node->GetNext();
            }
            while( node ) {
                AISTargetTrackPoint *ptrack_point = node->GetData();
                cc1->GetCanvasPointPix( ptrack_point->m_lat, ptrack_point->m_lon, &TrackPointB );

                dc.StrokeLine( TrackPointA, TrackPointB );

                node = node->GetNext();
                TrackPointA = TrackPointB;
            }

        }           // Draw tracks

So, it would be useful to capture a VDR trace of this type of target, just to understand the limitations of the technology.

Thanks
Dave

[PjotrC]

Using a stream of recorded AIS sentences I can see 3 types of tracks generated by OpenCPN:

- visual representation on the screen, as posted by Thomas
- gpx export of this, done from Target Detail screen
- a track recorded using VDM-VDO conversion for selected target

I compared these to hand-generated polygon linking AIS position reports.

The first 2 match exactly.

The third one differs a little, due to omitting some trackpoints.

I have no doubt the zigzag really occurs at the transmitting side.

[cagney]

It took some time, but now I have seen, and caught, the zig-zag again.....

Here is the track of the ship Las Palmas in Las Palmas Harbour Gran Canaria.



and the track-file.

AIS LAS PALMAS (255805591) 2015-10-26 19:48:17.gpx

Thomas

[bcn]

Thomas,

you don't have a VDR from this don't you?
As "Las Palmas" is traveling in and out from Las Palmas you might have an opportunity to catch it again...

Hubert