SBIR-STTR Award

Use of distributed computing in Navy systems is increasing. Distributed operating systems (DOS) provide the advantages of distributed computing to user applications. Requirements for real- time performance, reliability/fault tolerance, and security must be addressed by a DOS. Application of DOS technology is a complicated task. A knowledge-based assist will facilitate the use of distributed operating systems in Navy applications. Three technical objectives will demonstrate the feasibility of a knowledge-based DOS assistant: (1) develop a reference model for DOS usage; (2) show that DOS usage knowledge is amenable to knowledge-based techniques; and (3) show that generic and specific DOS knowledge can be integrated. In Phase I three objectives will be addressed by: developing a DOS reference model; defining the scope and structure of the DOS usage knowledge-base; developing rules for selection of real-time scheduling algorithms; and developing rule refinement techniques to ensure consistency and inheritance among different levels of knowledge. Phase I research will take advantage of existing results in knowledge-based design assistants from previous SBIR projects performed for the Navy and NASA.

Use of real-time computing in Navy systems is increasing, including the use of distributed computing. Real-time operating system services provide an Interface between application software and the hardware. Requirements for real-time performance, reliability/fault tolerance, and security must be addressed by a these services. Use of complex operating systems in applications having stringent requirements is itself a complex task. A knowledge-based assistant will facilitate the use of operating system services, especially distributed operating systems (DOS) services, In Navy applications. Phase II a prototype DOS assistant will be developed and demonstrated. The CLIPS expert system shell will be used as a framework for the DOS assistant. Knowledge will be acquired from internationally recognized experts in operating system usage and performance. The knowledge will be encoded as production rules. Demonstrations of the utility of the DOS assistant will be based on trial users developing realistic time-critical applications. Knowledge for a specific real-time operating system will be entered into the DOS assistant and the assistant will advise these trial users on alternative designs and implementations. Expert evaluation, simulation, and performance measurements will be used to determine the effectiveness of the advice offered by the DOS assistant.

Keywords:
REAL-TIME OPERATING SYSTEMS DISTRIBUTED OPERATING SYSTEMS EMBEDDED SYSTEMS