xoolive / rs1090

Decode Mode S, ADS-B and FLARM signals in Rust + Python

rs1090

rs1090 is a Rust library to decode Mode S, ADS-B and FLARM messages.

It takes its inspiration from the Python pyModeS library, and uses deku in order to decode binary data in a clean declarative way.

The project started as a fork of a very similar project called adsb-deku, but modules have been refactored to match pyModeS design, implementations extensively reviewed, simplified, corrected, and completed.

The directions ambitioned by rs1090 boil down to:

• improving the performance of Mode S decoding in Python;
• exporting trajectory data to cross-platform formats such as JSON or parquet;
• providing efficient multi-receiver Mode S decoding;
• serving real-time enriched trajectory data to external applications.

If you just want to decode ADS-B messages from your Raspberry and visualize the data on a map, you may want to stick to one of the dump0190 implementations.

The rs1090 library comes with a companion application decode1090 and a Python binding rs1090.

Performance

Benchmarking performed on the decoding of a gate-to-gate European flight:

• pyModeS in full Python mode;
• pyModeS with Cython compiled functions;
• rs1090 with Python bindings on a single core (for a fair comparison);
• rs1090 with Python bindings on many cores (default);
• full Rust rs1090 benchmark on many cores (default).

The Python script for benchmarking is in python/examples.
The Rust benchmark is executed with cargo bench.
Both scripts are run on an Intel(R) Core(TM) i7-10850H CPU @ 2.70GHz.

Note

The default out-of-the-box mode of rs1090 is an execution distributed on all your cores. This benchmark was performed on a regular laptop. It can be much faster on supercomputers, but considering that most laptops now have at least 4 cores, this benchmark yields the speed-up you should get on your own computer.

Installation

Run the following Cargo command in your project directory:

cargo add rs1090

Or add the following line to your Cargo.toml:

rs1090 = "1.0.0"  # check for the latest version, we are not there yet

Usage

use hexlit::hex;
use rs1090::prelude::*;

fn main() {
    let bytes: [u8; 14] = hex!("8c4841753a9a153237aef0f275be");
    // ADS-B decoding
    if let Ok((_, msg)) = Message::from_bytes((&bytes, 0)) {
        // JSON output
        let json = serde_json::to_string(&msg).expect("JSON error");
        println!("{}", json);
    }
}

See more examples in the crates/rs1090/examples folder.

Python bindings

You may install the bindings compiled for most Python versions with:

pip install rs1090

The library provides a single do-it-all function called decode():

>>> import rs1090
>>> rs1090.decode("8c4841753a9a153237aef0f275be")
{'df': '17',
 'icao24': '484175',
 'bds': '06',
 'NUCp': 7,
 'groundspeed': 17.0,
 'track': 92.8125,
 'parity': 'odd',
 'lat_cpr': 39195,
 'lon_cpr': 110320}

For large sets of messages in Python (e.g. what you can download through pyopensky):

>>> import rs1090
>>> rs1090.decode(msg_list, ts_list)  # includes CPR to position decoding
...
>>> rs1090.decode(msg_list, ts_list, reference=(lat0, lon0))  # useful for surface messages
...

For FLARM messages (also as batches):

>>> msg = "7bf2381040ccc7e2395ecaa28e033a655d47e1d91d0bf986e1b0"
>>> rs1090.flarm(msg, 1655279476, 43.61924, 5.11755)
{'timestamp': 1655279476,
 'reference_lat': 43.61924,
 'reference_lon': 5.11755,
 'icao24': '38f27b',
 'is_icao24': True,
 'actype': 'Glider',
 'latitude': 43.6812864,
 'longitude': 5.150585599999999,
 'geoaltitude': 970,
 'vertical_speed': 1.0,
 'groundspeed': 18.698261951315153,
 'track': 29.655457935479006,
 'no_track': False,
 'stealth': False,
 'gps': 129}

decode1090

Prebuilt binaries are available on the Releases page.
Usage is available with the help command.

decode1090 --help

rs1090's Issues

Possible updates required in patterns.json

Some counterintuitive examples to investigate:

• aircraft 4d24ca / 9H-VVI (among many others): registered in Malta, out of the Malta range (4d2000 to 4d23ff).
  Does the range need to be edited?
• aircraft c95000 / 4W-AAL: registered in Timor-Leste, undocumented range (Vanuatu stops at c903ff)
  Does Timor-Leste have a range?
• aircraft registrations in T2- are assigned to Tuvalu, no range in the patterns.json file. I have two occurrences T2-6965 (7d007c) and T2-6322 (7cd518), in the Australia range (7c0000 to 7fffff), mostly visible in Australia.
  What to think?

@espinielli any hint?

Quit gracefully without panicking

Just remove the .expect() and quit the loop

https://github.com/xoolive/rs1090/blob/78b4ec1ea499bab0ab112ecaed02f85317229d92/crates/rs1090/src/source/radarcape.rs#L20C1-L20C68

Timestamping issues

When the decoder runs for a longer time, it seems several messages for the same aircraft are issued with the same timestamp.

Suspicions (as of 21 June):

• restarting the samplicator program used to duplicate UDP feeds helps;
• messages do not necessarily reach the decoder in chronological order;
• we should also detect and merge messages seens by several decoders