ADisSAS: Adaptive Distributed Self Awareness System

We propose to build a platform to support the development of self aware processing systems. There are several contributions that the project makes. First, drawing upon our expertise in developing intelligent systems for system management, we define essential self awareness capabilities and define classes of core computations to support self aware processing. Secondly, we develop an interface that will allow human users to configure and direct a self awareness system. We define an architecture for deploying self awareness processes in an existing system using previous work on secure, self-configuring overlay networks done by ourselves and others. We provide an architecture that permits the coordination of multiple levels of self awareness based on our previous work on distributed workflow management. Finally, we explore how core self-awareness algorithms can structured to leverage multiple-core or reconfigurable microprocessors for acceleration. This will lay the foundation for the development of an appliance or board that could provide self awareness capabilities for many military and enterprise systems. Such a system could be developed as a commercial product during Phase III of this effort. At the conclusion of Phase II, the result will be a system that could be effectively deployed in a variety of important Military and Commercial civilian settings. For example, a version of our system could be deployed for managing a Future Combat Systems (FCS) based network. Alternatively, core self aware processing functions could be packaged in an appliance based upon COTS accelerator hardware and could provide enhanced self aware processing capabilities for blade or rack systems deployed in settings such as Air Operations Centers. We believe that our system, called ADisSAS (Adaptive Distributed Self Awareness System), could provide tremendous advances in system reliability and performance.