ISI’s Reconfigurable Computing group began investigating Field Programmable Gate Array computing in 1996, when FPGAs were tiny, homogeneous logic devices with limited I/O. As the lead integrator of DARPA’s Adaptive Computing Systems program, ISI aggregated emerging university and commercial prototype tools and devices, invented system architectures incorporating FPGAs, and applied them to real world signal and image processing applications, demonstrating stunning performance gains.
As architectures have become faster, larger, and heterogeneous, our research has evolved to consider operations in harsh environments (radiation and low-power), trusted design validation, and autonomous architectures.
Performance and reprogramability benefits of FPGAs have been well documented, yet serious programming challenges exist, especially for non-commercial environments and applications that add strict, unique requirements. The RC group specializes in this domain, resulting in novel application implementations and specialized tools to support deployment. Our focus:
Combining embedded processors with partial FPGA Placement and Routing Tools run-time reconfiguration has enabled a new class of introspective FPGA applications that can tune their performance during run-time. We’ve developed architectures that utilize schedulers or autonomous agents to develop situational awareness, and that reprogram the FPGA fabric during run-time based on dynamic or unanticipated events. Intelligent control targets cognitive radio, cognitive radar, robotics, and remote sensing applications.
ISI’s RC group develops tools that raise the level of abstraction and automate researched techniques for projects, enabling their participants to build quickly on previous successes. We’ve created run-time Application Programmer Interfaces, design environments for power optimization and radiation mitigation, and tools for the evaluation of trust and security of FPGA designs.
As an applied research lab, ISI proves and validates its research through demonstrations on real-world problems, utilizing full-scale applications. Our mapping techniques, including hardware software co-design and robust algorithm/architecture analysis, have enabled us to deliver real-time applications in cognitive radar, audio watermarking, radar, sonar, SAR ITR, wideband RF communications, HDTV encryption/decryption, and other vital areas.