Re: Testing resources


ross@...
 

Hi Paul,

Get the transmitter to send a known pseudo-random bit sequence that has good autocorrelation properties (eg a maximal length sequence from a linear feed back shift register: https://en.wikipedia.org/wiki/Linear_feedback_shift_register).

This is not encryption because the LFSR tap settings will be published and known to all (just like the "scrambler" used in K9NG/G3RUH 9k6 FSK modems).

Use a correlator at the receiver to demodulate this signal. The output of the correlator will tell you if you have any multipath echos (and what magnitude & delay they are, so that you can implement adaptive channel equalisation...). After channel equalisation, you can also use the correlator to determine S/N by measuring the difference between the correlation level and the noise level.
This technology has been around a long time eg every 2G GSM repeater sends repeated 64bit pseudorandom "training sequences" so that your mobile phone can sync to the repeater and adaptively equalise multipath and STANAG 4285 uses 80 symbols taken from a LSFR to achieve the same goal on HF. Adaptive channel equalisation is also absolutely vital for old fashioned dial up modems to operate 9k6 ~ 56k kb/sec in a 3 kHz voice channel. I ain't no patent lawyer, but I believe that there is plenty of published prior art from long enough ago for the relevant patents to have expired by now.


I have attached a spreadsheet (libre/open office .ods) and a few old screenshots. "channel equaliser.PNG" shows the TX bits, the raw RX bits, and the equalised RX bits. "impulse response1.PNG" shows the correlation function of the TX bits, and what comes out of the correlator at the simulated receiver. "impulse response2.PNG" shows the simulated receive correlator and what comes out of a correlator after the channel equaliser.

Basically, it sends the STANAG 4285 sequence over a bad channel (SNR is adjustable and the FIR is to simulate multipath), then it uses the autocorrelation properties of the received '4285 sequence to attempt to correct the channel impulse response (multipath). As you decrease the S/N, you will see that the channel equalisation is degraded.



73
Ross Whenmouth ZL2WRW

Join main@nw-digital-radio.groups.io to automatically receive all group messages.