Principles of Modern Radar - Volume 2 1891121537

766 CHAPTER 17 Advanced Processing Methods for Passive Bistatic Radar Systemsperformance with any of the algorithms due to the poor characteristics of the particularwaveform transmitted at this time. Even in this case, decreasing the detection thresholdthe target detection capability is not significantly enhanced. However, many false alarmsappear, making the target plot sequences more difficult to recognize and hence reducingthe capability to track potential targets [29].When using the ECA-B&S (Figure 17-11c), the completeness of the detected plotsequences is improved since additional plots are detected for a given target track. Thisgives a significant advantage especially for targets that result in only a two-plot sequencewhen operating with the ECA-B, since their tracks are otherwise likely to be lost. Moreover,an additional and quite complete sequence appears at about 60 km and –300 m/s that wasnot identified by the simpler algorithms.The ECA-B&S yields better detection capabilities with respect to the simpler versionsof the ECA and ECA-B. These advantages are obtained at the expense of a highercomputational load since additional cancellation stages are required to effectively removethe interference contributions and since the basis of the disturbance space is extended ateach stage, including the subspace spanned by the strongest target returns. However, theECA-B&S allows a reduction in the initial basis dimension of the cancellation algorithm,since the cancellation mask is adaptively updated. Thus, it avoids making the number ofDOF of the filter too large. In addition, the faster 2D-CCF evaluation algorithms describedin Section 17.2 might be used to speed up the processing at each stage.Finally, to further reduce computations, a suboptimal implementation of the ECA-B&S is introduced in [29]. The suboptimal implementation of the algorithm performs acancellation at each stage that involves only the small subspace spanned by the strongesttargets detected at the previous stage. This corresponds to an optimistic assumption thatthis subspace is orthogonal to the subspace spanned by the disturbance basis used in theprevious stage. The performed analysis of simulated and real data demonstrates thatthe two versions of the algorithm are quite comparable in terms of achievable performance,while the suboptimal version allows a dramatic decrease of the computationalburden.17.4 SIGNAL PROCESSING TECHNIQUESFOR REFERENCE SIGNAL CLEANINGAND RECONSTRUCTIONThe cancellation and the 2D-CCF evaluation algorithms presented in the previous sectionsassume the availability of a clean reference signal that corresponds to a highqualitycopy of the transmitted signal. Therefore, it might be necessary to perform sometransmitter-specific conditioning of the signal that depends on the exploited waveformof opportunity—for example, high-quality analog BP filtering, channel equalization, removalof unwanted structures in digital signals, and complete reconstruction from thereceived digital signal—before cancellation and 2D-CCF evaluation. Such conditioningapproaches are mainly intended for the reference signal aiming at improving its quality(with a benefit for the cancellation step) and the resulting AF (with reduced masking effectfor small targets).Except for the presence of thermal noise in the reference channel, the main degradationto the signal quality is typically related to the presence of multipath. Whereas the