Steganography through file rate modulation over HTTP - are there any ready-made solutions?

Question

Computer networks

“I came up with” a new way to hide the transfer of information from the curious and obliging to keep logs for 3 months, etc.
A sends B some useless file, a picture with a cat. The baud rate is modulated: slows down slightly to transmit "0", becomes normal to indicate "1". A third party can remove this information only if
1) it knows that such a transfer is in progress;
2) it is at the moment of transmission that it “listens” to the channel.
Surely, everything has long been invented. Question : have you ever met tools that can work with the speed of receiving / transmitting static files to hide additional files at the time of transfer. messages? Ideally, I would like to find a client part in JS and a server part in the form of an nginx module.
To protect against natural fluctuations in speed, it will most likely be necessary to simultaneously transfer two files, modulating their speed in opposite directions, inverting and adding at the end.

Answer 1

Question: have you ever met tools that can work with the speed of receiving / transmitting static files to hide additional files at the time of transfer. messages?

No, we haven't met, I even suspect why. I assume that this scheme is somehow workable only for a fixed implementation of protocol stacks, fixed hosts and fixed network parameters in between. Judge for yourself how this relates to reality.

To protect against natural fluctuations in speed, it will most likely be necessary to simultaneously transfer two files, modulating their speed in opposite directions, inverting and adding at the end.

What if the transfer duration is different for these files? For example, traffic with data of one file was transmitted over other physical links (etherchannel, L3 ECMP when interacting with different TCP connections). How to correctly compare (add) them? What does "simultaneously transmit" even mean? What will be the resulting payload data rate?
This idea seems fresh to me, but extremely impractical. Even if you encode the data in reserved TCP bits (3 bits per segment), you get a more applicable scheme.
In short, I think you will not find a ready-made implementation. If you implement the circuit yourself, then it will be very interesting what kind of performance you get and the range of parameters in which this circuit is operable.