Title of Presentation: Enabling Sensor Webs by Utilizing SWAMO for Autonomous Operations
Primary (Corresponding) Author: Kenneth Witt
Organization of Primary Author: WVHTC Foundation
Co-Authors: David Smithbauer, Jason Stanley, Al Underbrink, Mike Metheny, Dan Mandl, Vuong Ly
Abstract: The goal of Sensor Webs for Autonomous Mission Operations (SWAMO) is to shift the control of spacecraft missions to a collaborative, distributed set of intelligent agents versus a centrally controlled architecture. The network of intelligent agents will reduce management requirements by utilizing model-based system prediction and autonomic model/agent collaboration. These agents are distributed throughout the operational environment, which may include multiple spacecraft and operations centers, to monitor and manage spacecraft systems. This agent-based network forms a Sensor Web of agents via peer-to-peer collaboration and publish/subscribe message interaction. Some of the intelligent agents are mobile and thus will be able to traverse between on-orbit and ground-based systems. Other agents in the network are responsible for encapsulating system models to perform prediction of future behavior of the modeled subsystems and components to which they are assigned. Using situational awareness, the agents will be able to negotiate activities to self-optimize their subsystem or component. Within the web of agents there exist specialized agents responsible for separate mission control functional areas and tasking. Hierarchal abstraction of agent responsibilities creates high level agents that can break down larger system goals into subtasks, while low level agents respond to tasking associated with their specific capabilities. This allows for an ad hoc network, or Sensor Web, of agents to be assembled that is capable of autonomously executing mission control tasking. The basic SWAMO architecture consists of geographically distributed computing resources that host SWAMO agents. These resources may exist on the ground, in space, or on other flight platforms. The agents can seamlessly communicate via a publish/subscribe software bus architecture that spans ground and flight communication protocols. Each ground system and flight platform containing SWAMO agents and resources has the ability to run local software bus instances and connect to the global network instance via a bus-to-bus bridge. This overall bus architecture allows for greater autonomy within the SWAMO agent network by allowing access to and from flight applications to ground applications. The bus architecture promotes software reuse and the plug and play functionality. All agents within the network need only to build a single interface to the software bus instead of dedicated application-to-application interfaces. Agents can be added or removed from the network making it easy to add new Sensor Web capabilities and remove problematic or deprecated agent functions. SWAMO is an extension of the Model Based Operations (MBO) concept utilized for autonomous mission control of the Space Technology 5 (ST-5) constellation mission. SWAMO has been built. Since March 2008, SWAMO has been deployed on the US Naval Academy MidSTAR satellite as a demonstration platform illustrating real-time and planned autonomous management of constrained resources.