Principles of Modern Radar - Volume 2 1891121537

17.4 Signal Processing Techniques for Reference Signal Cleaning and Reconstruction 773the transmitter (e.g., band-pass filtering adopted to match the standard specifications,carrier frequency drift and oscillator phase noise, additional undesired modulations). Sucheffects are usually removed in the demodulation process, together with the transmissionchannel effects, since they introduce degradation in the extraction of the transmitted symbolsequence. In contrast to the case of the communication problem, where the informationcontent is of interest, the objective of the reconstruction process for a PBR is to obtainan exact copy of the transmitted signal that includes the above mentioned transmitterinducedsignal modifications. If these features are lost, their absence causes a decorrelationbetween the reconstructed signal and the signal received at the surveillance channel. Thismust be partially solved by postprocessing the ideal reconstructed signal with the aimto apply the same modifications introduced by the transmitter (in primis the shape ofthe passband filters), which can be either a priori known or estimated from the receivedsignal.In digital transmissions, an alternative and highly attractive approach is possible. Infact, the reference signal may be synthesized directly from the surveillance signal withoutrequiring a dedicated receiving channel. This can be obtained by demodulating the signalcollected at the surveillance channel to extract the sequence of transmitted symbols, whenit is possible to assume that the direct signal from the transmitter is by far the strongestcontribution received in the surveillance channel. As illustrated in the introduction, thisis often the case since the direct signal (despite the fact that it is received through thesidelobes) is typically much higher than the low-level signal reflected from any movingtarget. In these conditions, the reconstruction of the symbols sequence potentially yieldsonly a few errors even for SNR values that are not very high. In addition, as in theprevious case, proper equalization techniques and optimized demodulation and decodingschemes might be exploited for error correction and better recovery of the transmittedsignal. This approach in principle eliminates the effects of other signal contributionssuch as multipath and target echoes. Thus, a pure reference signal may be obtained byremodulating the extracted symbol sequence according to the standard specifications andinserting the additional transmitter-specific features, as discussed already.As an example, we report the experimental results obtained for a WiFi-based PBRusing the experimental setup described in [18, 21]. For this target detection experiment, thetransmitter of opportunity, a wireless access point (AP), was set up to roam for connecteddevices emitting the beacon signal at 1 ms intervals. A quasi-monostatic configuration wasadopted for the two antennas mounted one on top of the other. Two targets were present (acar and a running man) moving away from the antenna location at approximately 25 km/hand 20 km/h, respectively.Specifically Figures 17-16a–b show the 2D-CCF evaluated, before and after the disturbancecancellation stage, with an integration time of 0.4 sec. These figures were obtainedwith a pure reference signal directly collected at the output of the AP employing a directionalcoupler to gather a fraction of the transmitted signal in a dedicated receiving channel(reference channel). The 2D-CCF has been normalized to the nominal thermal noise powerlevel; the value at each map location represents the estimated SNR. The reported map hasbeen upper-saturated to a value of 50 dB so that the same color scale could be used for allthe reported results for a fair direct comparison (even after disturbance cancellation). Astrong peak appears in Figure 17-16a at zero range and zero velocity that corresponds tothe direct signal transmitted by the transmitting antenna and collected by the sidelobes ofthe surveillance antenna. Furthermore, all the 2D-CCF cut at zero velocity is characterizedby very high SNR values (well above the limiting color scale value of 50 dB) due to