Autonomic Performance Assurance for Multi-Processor Supervisory Control

Multi-processor computing systems are growing in capacity and usage. They encompass multiple, distributed implementations as well as heterogeneous, embedded computing architectures. The processing density enabled by such approaches holds promise for unmanned combat air vehicles (UCAVs) with their plethora of mission sensors and command and control processing requirements. However, the software and middleware required to effectively (and efficiently) harness multi-processor computing power is lacking. The team of Colorado Engineering, Inc. (CEI) and the University of Colorado, Colorado Springs (UCCS) proposes to design and develop an autonomic job scheduling system that addresses two areas of concern performance assurance and energy efficiency in distributed embedded multiprocessor systems for UCAVs. The effort will build on the Phase I program by developing a decentralized and scalable control system that supports automated job scheduling, performance assurance, fault tolerance, energy efficiency, and the migration of control (on-demand) into clouds with wireless networks in UCAVs. A prototype system will be tested in high performance embedded systems and cloud environments to validate performance and value. UCCS will draw on its research into autonomic computing, sustainable computing, and cloud computing, while CEI will leverage its experience with distributed, heterogeneous, high performance, multiprocessor embedded computing architectures during the proposed activities.