AFRL Analysis Tool Provides Over 10 Years of Real-Time Scheduling Solutions

  • Published
  • By Plans and Programs Directorate
  • AFRL/XP
Under AFRL's Dynamic Avionics Real-Time Scheduling (DARTS) program, Lockheed Martin has performed over 10 years of embedded hardware and software resource allocation trade studies for embedded information weapon systems such as the F-16, F/A-22, and the
F-35 Joint Strike Fighter (JSF). In particular, the DARTS schedulability analysis tool analyzes tasks allocated to available processors using rate monotonic scheduling techniques, ensuring the completion of all task deadlines under mission conditions.
The DARTS program introduced dynamic allocation technologies into the domain of embedded, resource-constrained real-time avionics systems, a subject area in which scientists rarely use these technologies. DARTS has supported successful in-context JSF demonstrations of dynamic binding and scheduling technology, analysis of F/A-22 avionics stability problems, and initial schedulability tests for production
F-35 mission systems designs. Using DARTS' distributed, multiprocessor resource management capabilities as a foundation, the Defense Advanced Research Projects Agency's Adaptive Software Technology Demonstration project added features to assess embedded applications with time-varying loads.
DARTS' numerous technology benefits are applicable throughout the system design and implementation process. Releasing system designers from the requirement of binding tasks to processors too early in the design process--long before the formulation of realistic resource estimates--allows the designers to focus on solving domain problems. Likewise, delaying the time at which designers must make task-to-processor binding decisions allows the system to automatically cope with the additional task interactions that are not in the original design. Dynamic binding simplifies subsequent system upgrades by building into the platform the ability to relocate tasks as other tasks grow or diminish in priority.
The DARTS schedulability analysis tool has continued to evolve. It forms the basis for DARTS Architect, which offers the capability to rapidly and accurately perform systemwide allocation trade studies by integrating models of network throughput, node-level rate monotonic central processing unit utilization, hardware memory, and bus bandwidths. The DARTS program's resulting products continue to lead in the demand for advanced middleware and scheduling theory for embedded information systems.