This is a Mode S multilateration server that is designed to operate with clients that do not have synchronized clocks.
It uses ADS-B aircraft that are transmitting DF17 extended squitter position messages as reference beacons and uses the relative arrival times of those messages to model the clock characteristics of each receiver.
Then it does multilateration of aircraft that are transmitting only Mode S using the same receivers.
See commits or a diff for details.
It is important that you read this section before using or modifying the server!
The server code is licensed under the Affero GPL v3. This license is similar to the GPL v3, but it has an additional requirement that you must provide source code to users who access the server over a network.
So if you are planning to operate a copy of this server, you must release any modifications you make to the source code to your users, even if you wouldn't normally distribute it.
If you are not willing to distribute your changes, you have three options:
- Contact the copyright holder (Oliver) to discuss a separate license for the server code; or
- Don't allow anyone else to connect to your server, i.e. run only your own receivers; or
- Don't use this server as a basis for your work at all.
The server will automatically provide details of the AGPL license and a link to the server code, to each client that connects. This is configured in mlat/config.py. If you make modifications, the suggested process is:
- Put the modified source code somewhere public (github may be simplest).
- Update the URL configured in mlat/config.py to point to your modified code.
None of this requires that you make your server publically accessible. If you want to run a private server with a closed user group, that's fine. But you must still make the source code for your modified server available to your users, and they may redistribute it further if they wish.
- Python 3.4 or later. You need the asyncio module which was introduced in 3.4.
- Numpy and Scipy
- python-graph-core (https://github.com/pmatiello/python-graph)
- pykalman (https://github.com/pykalman/pykalman)
- optionally, objgraph (https://mg.pov.lt/objgraph/) for leak checking
- gcc
- uvloop, ujson, Cython
apt install python3-pip python3 python3-venv gcc
VENV=/opt/mlat-python-venv
rm -rf $VENV
python3 -m venv $VENV
source $VENV/bin/activate
pip3 install -U pip
pip3 install numpy scipy pykalman python-graph-core uvloop ujson Cython
After every code update, recompile the Cython stuff:
source $VENV/bin/activate
cd /opt/mlat-server
python3 setup.py build_ext --inplace
Starting mlat server:
export MKL_NUM_THREADS=1
export NUMEXPR_NUM_THREADS=1
export OMP_NUM_THREADS=1
$VENV/bin/python3 /opt/mlat-server/mlat-server
(example has git directory cloned into /opt/mlat-server)
For an example service file see systemd-service.example
It's all poorly documented and you need to understand quite a bit of the underlying mathematics of multilateration to make sense of it. Don't expect to just fire this up and have it all work perfectly first time. You will have to hack on the code.
$ mlat-server --help
You need a bunch of receivers running mlat-client: https://github.com/wiedehopf/mlat-client The original version by mutability will also work but the wiedehopf client has some changes that are useful. (https://github.com/mutability/mlat-client)
Results get passed back to the clients that contributed to the positions. You can also emit all positions to a local feed, see the command-line help.