Principles of Modern Radar - Volume 2 1891121537

17.6 Multichannel Processing for Detection Performance Improvement 813uncanceled, but their removal can be obtained with the adaptive temporal filter. A referencebeam, orthogonal to the surveillance one, has to be digitally synthesized and yields thereference signal to be used by the adaptive temporal filter. Finally, the 2D-CCF is evaluatedand the CFAR detection map is built.Sidelobe Canceller An alternative approach can be obtained using the sidelobe cancellerconcept [63]. For instance, with reference to the specific array configuration previouslyconsidered (see Figure 17-41b), we might exploit seven of the eight elements of the arrayto form the quiescent beam and the central element as an omnidirectional retrodirectivebeam. Thus, the resulting weights vector for the sidelobe canceller is given byw H SLC (θ 0) = s H θ (θ 0) T H ( TMT H ) −1T (17.48)where the matrix T is defined as[s∗(1)θ (θ 0 ) s ∗(2)θ (θ 0 ) ··· s ∗(7)]θ (θ 0 ) 0T =0 0 ··· 0 s ∗(8)θ (θ 0 )(17.49)where s (i)θ (θ 0 ) is the i-th element of the target steering vector (i = 8 refers to the centralelement). Notice that in this case a 2 × 2 matrix in place of a M × M matrix has to beinverted. Even in this case, the array pattern strongly attenuates only the signals comingfrom DOA close to the direct signal; therefore, the multipath reflections have to be canceledby the adaptive temporal filter.To show the effectiveness of the considered spatial cancellation techniques, a performancecomparison is presented below for a simulated FM-based PBR scenario exploitingthe array configuration in Figure 17-41b. The results have been averaged over several runs.Aiming at assessing the detection performance of the considered techniques, PBR scenarioswith a relatively large number of targets have been considered. A total of 16 targetshave been injected in the simulated data with target/receiver distance ranging from 10 to100 km, DOA between 150 ◦ and 180 ◦ , bistatic velocities in the range [–150,200] m/s, andSNR between –35 and –55 dB. As expected, in the absence of multipath/clutter contributions,the three adaptive approaches yield comparable performance allowing detectionof 13 of 16 targets. However, when such contributions are considered in the simulateddata, the optimum filter experiences some detection loss. Specifically, it fails to detecttargets with DOA close to the DOA of the multipath returns. In this case, the principaleigenvalue and the sidelobe canceller approach yield more robust detection performance.Figure 17-44 reports the detection results obtained for a simulated scenario containingseveral stationary scatterers with clutter-to-noise ratio between 25 and 30 dB.As is apparent, the optimum filter (Figure 17-44a) yields an effective removal of allthe undesired disturbance contributions that also involves many of the injected targets.In contrast, using a reduced number of DOF, as required by the principal eigenvalueapproach (Figure 17-44b) and the sidelobe canceller (Figure 17-44c), prevents undesiredtarget cancellation at the DOA of the strong multipath echoes, thus yielding remarkabledetection performance.The many possibilities opened by exploiting adaptive array of antennas for PBRsignal processing offer a wide range of possibilities for both detection (where the previousexample results have been shown) and for DOA estimation, where the antenna array mightallow for increased accuracy and robustness.