01 / Communications research
Encode
Map information into a controlled signal.
Learned coding is one candidate alongside conventional baselines. Cryptographic authentication and encryption remain separate responsibilities.
Communications research
A radio link you can understand end to end.
Neural-assisted coding and receivers meet an explicit physical link budget. Explore the path from a message to a waveform and back.
Design and research · public preview
DarkRadio / system sketch
How it works
Follow the system.
One part at a time.
Select a part to see its role, the decisions behind it, and what still needs evidence.
02 / Communications research
Transmit
Account for bandwidth, antennas and permitted spectral power.
Impulse UWB and lower-frequency narrowband are distinct research directions. Their power, timing and propagation assumptions cannot be exchanged.
03 / Communications research
Recover
Estimate the channel and combine available signal.
Channel estimation, equalization and multipath processing are possible roles for neural methods. Clock jitter and receiver noise still constrain the link.
04 / Communications research
Compare
Measure decoding against a conventional baseline.
Held-out channels, payload error rates and total processing cost are needed before a learned receiver has a demonstrated advantage.
Design principles
What guides the work.
Physical accounting
Keep propagation, body loss, antennas and allowed bandwidth visible.
Independent building blocks
The communications link does not depend on an optical processor being built first.
Exact packet boundaries
Neural processing does not replace packet authorization or key custody.
Project status
Know where
the work stands.
Architecture and research. No measured range, payload rate, detection probability or complete link result has been established.
Source and project notesThe source repository requires authorized access. This public page contains a curated explanation and original schematic illustrations.