How do it faster

[weidaxu1988]

Hi,
I am trying out this project, and it works nicely. But my case is time sensitive, I use iphone and my nexus phone as transmitter, raspberry pi as receiver, and I send 32 bytes once a time. It costs around 800ms for both prefix audible mode and ultrasonic-experimental mode.

Is there any way that I can do with config file to make it faster?

PS: distance is not a issue, I can make it close enough. ^^

Any suggestion from you will really help me.

[brian-armstrong]

This will ultimately come down to which profile you use. You could, as a starting point, try out some of the other profiles just to see if they get closer to what you want. The profile names are just the top-level keys in quiet-profiles.json

[scream7]

Hi @brian-armstrong
I think @weidaxu1988 means to adjust parameters of default profiles, not only apply the default settings.
And I also faced to this situation.

[brian-armstrong]

One option would be to try out https://quiet.github.io/quiet-profile-lab though unfortunately it hasn't been updated with all the newest features.

In general, changing the modulation and samples per symbol will have the greatest effect on the speed

[weidaxu1988]

@scream7 yes, @brian-armstrong I tried all prefixed method, the result seems the same. I also change the frame_lenght, checksum method, but not quiet good. I also realized that send 32 bytes need around 800ms, can I reduce length from the sender? 32Byte is all I need.

[brian-armstrong]

Oh, I bet you're running into the audio latency. There is a way to shorten this.

You'll want to change numBuffers and bufferLength on QMReceiverConfig and QMTransmitterConfig.

You can see the default values here https://github.com/quiet/QuietModemKit/blob/master/QuietModemKit/QMReceiverConfig.m . You'll want bufferLength to be a power of 2.

Presently there are 3 buffers of 4096 samples each on transmitter and receiver side. 2 * (3 * 4096) / 44100. = 557 ms. I bet that's most of what you're seeing. There are other latencies that could be improved but this will be the largest one.

Reducing the buffers will make it more likely that the app stutters and doesn't receive correctly, so you'll have to experiment. This is just an area I haven't optimized yet :)