This research develops new fast adaptive filtering algorithms and structures, which are characterized by a low computational complexity, and is applying them to a broad class of applications, including modern communications systems and advanced sensor array processing. For the communications application, this study demonstrates that such adaptive signal processing can effectively mitigate interference in satellite and terrestrial CDMA spread spectrum wireless communications as well as provide cross-talk cancellation in dense WDM fiberoptic links. The benefits of this new technology include power savings and increased data throughput, leading to newer generations of receiver processing hardware that are being developed from this study. In the sensor array processing application, it is demonstrated that low-complexity, reduced-dimension adaptive CFAR detectors are capable of outperforming all existing detectors for modern radar systems. The technology developed from this research permits space-time adaptive processing on platforms which are limited in terms of size, weight and power; such as airborne and space-segment surveillance systems. The key underlying innovation in all of these applications is the use of transform domain adaptive signal processing utilizing the cross-spectral metric and low-complexity filtering structures. This metric permits a lower-order adaptive filter to converge to a solution which is either identical or very close to the optimal solution. Since this filter has a smaller number of adaptive weights, it converges faster and requires a lower computational complexity. This new technology represents a generalization of multiresolution and time-frequency basis representations. The full-dimensional transform domain representation (realized using a filterbank, DFT/DCT, or wavelet) is then compressed to keep only the most important information as selected by the cross-spectral metric.
