Array Processing in the Face of Nonidealities

Real-world sensor arrays are typically composed of elements with individual directional beampatterns and are subject to mutual coupling, cross-polarization effects as well as mounting platform reflections. Errors in the array elements’ positions are also common in sensor arrays built in practice. Such nonidealities need to be taken into account for optimal array signal processing and in finding related performance bounds. Moreover, problems related to beam-steering and cancellation of the signal-of-interest in beamforming applications may be prevented. Otherwise, an array processor may experience a significant performance degradation. In this chapter we provide techniques that allow the practitioner to acquire the steering vector model of real-world sensor arrays so that various nonidealities are taken into account. Consequently, array processing algorithms may avoid performance losses caused by array modeling errors. These techniques include model-based calibration and auto-calibration methods, array interpolation, as well as the wavefield modeling principle or manifold separation technique. Robust methods are also briefly considered since they are useful when the array nonidealities are not described by the employed steering vector model. Extensive array processing examples related to direction-finding and beamforming are included demonstrating that optimal or close-to optimal performance may be achieved despite the array nonidealities