The RTOS (Real-Time Operating System) includes an extensible and configurable kernel such that common kernel software can be incorporated within embedded systems in the space segment and within the ground segment to the greatest extent possible. Current work on RT-Mach, RTEMS, and many other open operating systems which support real-time scheduling can be used to provide this functionality. These operating systems can be used to easily integrate the other major functional layers of this architecture with a common kernel layer that provides standardized abstractions of the processors and hardware interfaces incorporated in the overall system.
The generality of an RTOS in comparison to an application specific executive does mean that greater overhead is placed upon task management for a given application, but it also means greater flexibility, quicker development, and better scalability and ability to support the automation layers of the overall system such as detection and rule-based reaction. For example, the ability to meet deadlines within a real-time system with rule-based inferencing can be greatly simplified by an RTOS scheduler which provides methods to balance rule-based inferencing with other system processing and critical processing with hard deadlines. The reliability of real-time systems incorporating complicated inferencing and planning algorithms can be greatly increased with proven scheduling algorithms [8,9].