Early results - The approximate locations of contributors are shown in red blobs (Hope Obj doesn’t mind me sharing this 
)
This is the present coverage map of the MLAT Beta
MLAT is short for multilateration - about half the aircraft don’t transmit positional information - these are the planes you see listed in white on the DUMP1090 web page. What multilateration does is try to work out where the plane is by calculations based on the time it takes for the signal from the plane to reach three receivers, with more confidence in accuracy of fix if more receivers are used in the calculation. The use of more receivers in the coverage area also allows fixes to be calculated for planes at lower altitudes.
The remarkable thing about this is that it shouldn’t work - the USB connected dongle isn’t supposed to handle the signals in a predictable enough way for the calculations to be done … but work it does.
If you have a Pi & dongle based receiver with a range to reach two or three of the the present stations participating, and want to take part - send Obj an email (address is on his post about 5 posts up. Obj has an MLAT client-package for you to install on your pi.
Very nice work!
Cheers!
LitterBug
Hi,
Is this advanced enough so that others can join to extend the zone?
I am too far south to get with you, (my northern reach is about troyes/nancy), but if any other france/switzerland users this would nicely add up to your current coverage.
It’s up to Obj if he wants to set up another zone since he is is providing the compute and network resource doing the processing (I think it would be interesting to do)
To get it working you need a minimum of 3 stations (four or more is better) with a good range widely spaced enough that they produce different data but close enough that they have a good overlap of range.
It’s probably easier to extend what is now working and increase the density of receivers in the covered area (would make more lower flying aircraft detectable)
–
I would like to know if it might be possible to de-centralise the ‘analysis’ processing to say a geographic area (say a country), and have each client able to feed to three or four decentralised servers (think of a station in the Alps might feed to Switzerland, France, Germany & Italy for analysis - then bring the results together for publication.
I wouldn’t mind providing / hosting a high powered machine like a Pi2 to provide the regional analysis (as long as the amount of traffic didn’t upset my Internet provider)
Early days yet, but I really like that we are able to do this using recieving stations that cost under $100 that cost pennies to run when some tracking organisations are spending maybe ten times as much to do a similar job.
Interesting
An MLAT zone FR-CH-DE would be great for me. However you are right the network bandwith can be an issue here.
1/How much upload does one need to feed the mlat analyzer?
2/ How much processing power is required to run the analyzer?
3/ How does one who uploads mlat data gets the positions back for its own use/display?
Regards,
Yes this is somewhere on my todo list. The nature of the problem lends itself well to geographic separation like this.
The current server can probably take another half-dozen receivers before CPU starts getting tight.
Receiver upload is something like 2-4kB/s depending on message rate. This is already compressed, but it’d be possible to reduce it further with some more work (being more selective about which messages to send)
The current server runs on a 1-vCPU rackspace cloud server at 40-60% with the current set of receivers (about a dozen receivers tracking 20-40 aircraft). There is a lot of optimization possible there. Going beyond 1 CPU gets trickier due to python limitations (its threading model is not great for CPU-bound apps), probably the next step after optimizing the server code is geographic separation.
The server returns position results directly to the clients that contributed to the result via the same connection the clients used to send data. The clients turn that into a Basestation-format feed locally.
Let me know when you want to expand to Silicon Valley.
In another month our server in the closet will be going from dual processor to dual quad cores and a lot more memory, so we’d probably be able to host computation (depending on total bandwidth in and out). We also have our own Stratum 1 NTP server.
bob k6rtm
Can we have a parameter for the client so we can choose to receive the back-chat or not - save resource if it’s not used.
If you set the basestation port to 0 then it’ll do this.
If you want an additional test region, I would be able to beta test the MLAT software. I’m in NC, USA. If only there were some other folks in NC. Cough, cough…
FA operates a lot of receivers in East Texas that could be ideal for a zone. We also have massive amounts of computing and hosting resources. Ideally we would bake this directly into PiAware so it’s available by default for everyone.
Thumbs up for this one :-bd
Would be such a big improvement for FA 
to Obj & Co, those who want to prove it’s possible and all the early testers …
Kind regards from EBBR/BRU
I am between Münster and Bremen in north west part of Germany. A little bit too far to the stations I guess.
If there would be more stations I like to test it too.
+1
Yes, Yes. Yes!
Obj has done so much work on developing a prototype that shows what shouldn’t work can produce good results - he’s shown great interest and talent with what he’s done so far. (Keep up the good work)
LudgerBoergerding - send an email to Obj - Your location is ideal to extend coverage eastwards
And fully agree on all the work done to get it this far!
My total range (red area, rings are 50nm each) of around 200nm is now pretty much covered by MLAT (darker area) too ! The east could be extended a bit more so LudgerBoergerding: go for it! 8)
It’s cool finally to be able to see F16’s chasing each other.
/paul
I wrote an email to obj. My current coverage look like this:
https://www.dropbox.com/s/bikkb47gy51x3fy/Screenshot_2015-04-02-11-32-43_1.jpg?dl=0
I still need to optimise my antenna. I want to build it by myself.