Full Chain Benchmarking for Open Architecture Airborne ISR Systems: A Case Study for GMTI Radar Applications

Abstract : As the airborne ISR application space evolves, the quantities of data acquired by remote sensing systems such as radar, electro-optical, and infrared systems are growing larger, and advanced algorithms are imposing more challenging computational requirements for real-time processing. While the difficulties in processing sensor data in real-time is the topic of extensive research, the rapidly shifting technology and application complexity has led to pronounced system lifecycle challenges, including the constant threat of technology obsolescence and unsustainable maintenance costs. One way for Government programs to address this reality economically is to shift the ISR system acquisition strategy to facilitate the timely, cost-effective insertion and upgrade of technologies, through the utilization of an open architecture (OA) approach to system design standards for application ready processors (ARPs). OA design leverages industry-standard hardware and middleware, thus engaging a broader development community and lowering barriers for third-party application development. For this approach to succeed without sacrificing functional capabilities and real-time performance, effective benchmarks are necessary to ensure that an ARP system can meet the mission constraints and performance requirements of real-world applications. This work investigates the measurement of real-time performance of commodity high-speed processing solutions and middleware for airborne systems using OA composite benchmarks, i.e., benchmarks that characterize computational performance of the system as a whole, while also validating OA principles. For ISR systems, processing performance must often be counterbalanced by size, weight and power (SWaP) constraints that often necessitate application-specific configurations (e.g., mapping and scheduling) for system-level optimization.