Innovative Approaches for Embedded Real-Time UHF/VHF SAR Image Formation

There is a critical need for an all weather capability to detect and classify military targets concealed in foliage and/or camouflage. Recent advances in high resolutions synthetic aperture radar (SAR) technology, operating at low frequencies (VHF-UHF), have demonstrated significant capability for detection and classification of such targets. However, one of the primary difficulties is the significant amount of processing throughput and memory required to realize real-time operation. Integrated Sensors, Inc. (ISI) proposes an implementation which leverages the high throughput of various specialized DSP devices by integrating them into a highly efficient heterogeneous parallel processor optimized for the FOPEN SAR function. Algorithm alternatives, partitioning, and parametric trades will be evaluated with the purpose of achieving maximum algorithm-hardware synergy. Improvements in critical processing performance areas will be demonstrated using representative hardware components, such as SHARP's BDSP 9124block floating point device, which will be integrated into ISI's Mercury based heterogeneous test bed. By making judicious algorithm tradeoffs and carefully matching specific processing resources to different aspects of the problem, one can reasonably expect an order of magnitude improvement in performance. A strawman design for a Phase II real-time heterogeneous processor will also be developed.