Real-time adaptive control of knowledge based avionics tasks

Advanced decision-making capabilities are being developed to aid the pilots of the next generation of tactical fighters. Due to the limited processing resources available in an avionics suite, efforts have focused on developing a distributed fault-tolerant software architecture that permits the real-time prioritization and scheduling of these tasks. The authors outline the design details of an architecture under development to meet these performance requirements. The system has been tested with a threat-avoidance system, implemented on a testbed of five internetted LISP workstations, to evaluate overall system capabilities including scheduling, task operations, and database accesses. It has a simulation cycle of 50 ms and synchronization between distributed nodes can be achieved within 2 ms. This test case has nine knowledge tasks, one of which is defined as a simulation cycle that drives the test case. This system has been evaluated with the current trace capabilities and runs with a peak of 16 task instances active at any time