GNGTS 2015 - Atti del 34° Convegno Nazionale

GNGTS 2015 S essione 3.3 161 In the following, the reconfigurable system will be briefly described in section 2, then the algorithm for the reconfiguration of the integration times is exposed in section 3. Some preliminary experimental results achieved in the field are then exposed in section 4. Conclusions follow in section 5. The reconfigurable system. A prototypal reconfigurable GPR system (Persico and Prisco, 2008) has been implemented within the research project AITEACH, as said. Several tests on real sites in Italy and abroad (Norway and Malta) have been performed within a Ph.D. course handled within a collaboration by the Department of Earth Science of the University of Bari and the Institute for Archaeological Heritage of the Italian Research National Council (IBAM- CNR). Fig. 1 (in the left hand panel) offers an image of the system, that allows to reconfigure vs. the frequency the length of the antennas, the power radiated at each frequency and the integration time of the received harmonic tones. The length of the equivalent antennas of the prototypal reconfigurable GPR system can be modified by using special switches that can make the arms “longer” or “shorter” as shown in Fig. 1 (in the right hand panel). The arms of the antennas can be electronically reduced by means of two internal switches ( is ) and two external switches ( es ). When both the is and the es are switched on, the antennas are “long”, when the is are switched on and the es are switched off, the antennas are “medium”, if both the is and the es are switched off, the antennas are “short”. Of course, the switches of the equivalent antennas are sycrhonised so that the the lengths of the transmitting and receiving antenna are constantly equal to each other. The reconfigurable antennas are comprehensively sized about 80x45 cm times 25 cm for the height. The central unit is attached to the antennas, so that no external cable is needed to connect them to each other. The frequency interval that the system can sweep ranges from 50 MHz to 1 GHz with a frequency step of 2.5 MHz or (optionally) of 5 MHz. In the second case, the measure can be performed in a slightly faster way, but the risk of aliasing in frequency domain increases (Persico, 2014). The three central frequencies of the equivalent antennas can be evaluated from the data. In previous works, as (Persico et al. , 2014; Matera et al ., 2015), some possibilities related to the variable length of the arms of the antennas were shown and discussed. Here, we focus on the reconfiguration of the integration times whereas, at the moment, the possibility of reconfiguring the radiated power vs. the frequency has not been deeply investigated yet. The reconfiguration of the integration times. The algorithm for the reconfiguration works analyzing a first calibration Bscan, possibly coinciding with the first Bscan of interest for the prospecting at hand. This preliminary Bscan is exploited in order to evaluate and quantify the presence of meaningful (if any) interferences. This aim is pursued as follows: for all the traces and for each tone within the trace, N samples of the in phase and in-quadrature ( I and Q ) components are retrieved and stored, by means of the customary heterodyne demodulation receiving chain of any stepped frequency system. Indeed, this is not a procedure conceived on purpose for the scopes of this paper, but is the normal praxis customarily adopted in stepped frequency systems. Usually, however, one accepts as I and Q components the algebraic average value of the N I and Q samples gathered at that frequency in that position. Here, we propose to exploit another information related to these N samples, namely their variance. In particular, in Fig. 2 we show the results of an experimental test performed in Florence by the Florence Engineering s.r.l., relatively close to a repeater of FM broadcast radio transmissions. In abscissa there are the subsequent I and Q samples frequency per frequency (but indexed after their sequential number). In correspondence of the frequencies not much disturbed, the samples are substantially equal to each other within the tone, thus resulting in a step like diagram. However, when the tones are disturbed by external electromagnetic devices, then the behaviour of the samples within the tone is quite variable and noise-like. This has suggested to deal the in-phase and in-quadrature components as the samples of two random variables. Since the algebraic average is an estimation of the statistical average, we can