Principles of Modern Radar - Volume 2 1891121537

17.3 Direct Signal and Multipath/Clutter Cancellation Techniques 761with dimension K × K , which in this case corresponds to O[N b K 2 + K 2 logK ] complexproducts. Clearly this computation needs to be repeated on each batch, thus yieldingonly a limited increase in the computational burden (about the same order of magnitudein terms of complex products), namely, O[NK 2 + bK 2 log K ] in place of O[NK 2 +K 2 log K ]. However, at each batch the dimension of the data can be reduced by afactor b, thus reducing the dynamic storage requirement of the system. In addition, thecancellation processing can run parallel with the data acquisition since it can be startedafter the first batch has been received. This typically makes the implementation easierand faster.2. The size N b of the batch sets the time extent over which the adaptive filter parametersare estimated. Reducing N b within certain limits increases the adaptivity loss whenoperating in a stationary environment. However, it allows a certain degree of adaptationto the slowly varying characteristics of typical transmissions, thus making the systemmore robust.3. By operating with a reduced temporal extent, T b , the Doppler resolution capability ofthe cancellation filter is degraded with respect to the ECA approach operating over thewhole T int period. This results in a wider notch in the Doppler dimension that in turnremoves the potential request for additional DOF to cope with the spectral dispersionof the interfering signals that would require a higher computational load.Figures 17-7c–d show the 2D-CCF and the detection results obtained for the same datafile of Figures 17-7a–b after the application of the ECA-B for b = 10 and K = 50. Thewidth of the notch in the Doppler dimension is significantly increased in Figure 17-7c withrespect to Figure 17-7a, which was obtained with ECA operating with the same number ofDOF. This yields a better cancellation of the disturbance, thus allowing the strongest targetsto be easily recognized together with their sidelobe structures. The detection performancehas been clearly improved since potential targets are now detected at bistatic ranges greaterthan 120 km.Detecting the weakest targets is limited no longer by residuals of the cancellationprocess but now by the masking effect of the sidelobes of the strongest targets. So it isuseless to keep increasing the dimensionality of the disturbance subspace; the DOF savedby using the ECA-B approach can be devoted to counteract this additional limitation. Anad hoc algorithm is presented in [29] and briefly summarized in the following section.17.3.3 ECA Batches and Stages (ECA-B&S)A complete processing algorithm is presented, based on the ECA-B approach, that allows:a. A preliminary cancellation of the main disturbance contributionsb. A cancellation performance refinement by the adaptive definition of an extended cancellationmaskc. The removal of the strongest targets, thus yielding the detection of the weakest onesThe idea of using an iterative cancellation approach is preliminary addressed in[30, 31]. Further, a sequential cancellation algorithm is presented in [34] that exploits avariable number of iterations adaptively identified from the measured data. The approachwas designed to prioritize cancellation of the largest disturbance until a desired, predeterminedcancellation level was reached. While the ordering strategy and stopping condition