2G, 2G frequency hopping, 4G spectrum Fig. 3 shows the types of communication the program can recognise. These are obtained by keeping only maximum values for each frequency during the analysis. Instant values are not relevant in this part of identification because for traffic channels, information may not always be transmitted, as for beacons. To be able to differentiate those three, one more vector is considered in order to store the values of the central frequencies shifted at 100 kHz, which is the end of the GSM channel. Based on it and imposing an empirical threshold, in our case 4 dB between two adjacent channels, the program makes the difference among the types of spectra. Also, to differentiate whether a 2G channel is beacon or traffic, instant values were taken into consideration, a different threshold than the one mentioned above, and a possible error. For a channel to be a beacon [4], it means that it transmits continuously and every sample must be above the noise level. If the instant value is over the threshold (quite bigger than the noise power, because it can fluctuate), it is counted and compared to the total number of comparisons made on that frequency. In this analysis, we noticed that beacon channels never had all samples over threshold, so a small deviation is empirically set which recognizes the type in more than 85% of cases. Another way to differentiate the beacon from traffic is to analyse the FCCH (Frequency Control Channel) [7], as in Fig. 4. Only beacons have it and can be seen in the spectrum with a small spike close to the central frequency of each channel. This solution is not chosen, because it is based on resolution, meaning that if the FFT size is too small, this channel cannot be noticed properly and that could affect the decision.