GMSK comb Spectrum Sensing for Machine Learning

<p>These files contains measurement results achievied in the following scenario:<br> Single PC with GNU Radio and connected USRP transmits GMSK comb signal (6x GMSK signal at center frequncies 2097.5 MHz, 2098.5 MHz, 2099.5 MHz,&nbsp;2100.5 MHz, 2101.5 MHz, and 2102.5 MHz) with different amplifier gain.<br> Single PC with GNU Radio and connected USRP receives signal with central frequency 2100 MHz and bandwidth 40 MHz (treated like 40 x 1 MHz channels). It uses local osciallator offset equal 10 MHz and due to non-linear characteristics of amplifiers and filters in USRP the extreme 12 (on both sides) are removed. However to create (this) dataset only 6 channels where signal was transmitted were included.<br> <br> Received is set in three different possitions (similar height and distance from transmitter)&nbsp;- creating 3 files -&nbsp;and collects data to detect presence of transmitted signal. Data that can be found in the files is stored in CSV format to be easly analyzed in ML models.&nbsp;</p> <p>Data collected at during this experiment contains:<br> six columns of average received power in the analyzed channels (in dBm)<br> six columns of autocorrelation function kurtosis in the analyzed channels (in linear scale)<br> six columns of autocorrelation function skewness in the analyzed channels (in linear scale)<br> one column with information about signal transmission (label; 0 - noise, 1 - signal transmitted)<br> one column with transmitter gain (in dB; set to -100 in case of lack of transmission)</p>