/sci/ - Science & Math

[Anonymous No.3610152]

Hello /sci/

are there any fellow electrical engineers/signal processing/telecommunications geeks out there?

I'm trying to implement a FSK-demodulator in software and have some questions regarding the general approach/pitch detection algorithms

[Anonymous No.3610167]

What seems to be your problem? Engineering physics type here, *might* be a little help :P

[Anonymous No.3610250]

Alright, so this here is my signal. 16 bits with 10kHz = 0 and 12kHz = 1

It is periodically sent (about 5 times per second, see second pic in next post).

My receiver records this transmission into a 1sec-buffer and now I have to decode that stuff back to ones and zeros.

Wat do? Autocorrelation? FFT? Goertzel algorithm? It doesn't have to be very quick, but FFT seems to be too slow
Everything has to be done in software(Java)

[Anonymous No.3610256]

[Anonymous No.3610286]

>>3610250
>Everything has to be done in software(Java)
>Java

I feel very sorry for you.

[Anonymous No.3610334]

I take it you can't assume source-synchronous transmission?

[Anonymous No.3610378]

>>3610286

Thanks bro

>>3610334

Correct

[Anonymous No.3610393]

A quick google suggests bandpass filter, then take RMS.

[Anonymous No.3610458]

>>3610393

Thanks. Sadly digital filters won't work since they need a few samples settling time. The sample rate will be around 44.1kHz

[Anonymous No.3610479]

It looks like you have a rather weird waveform there. Does the amplitude of the transmitted signal increase and decrease in the shape of a triangle wave for each transmitted bit???

Or is that just aliasing in your image?

[Anonymous No.3610544]

>>3610479

Yeah, I've reduced the attack and delay time for each bit on purpose for clearness and avoiding interference

Here's a version without

[Anonymous No.3610557]

matched receiver