Data Fusion for Space Situational Awareness

Scientists demonstrate data fusion and decision support tools for satellite threat analysts.

Satellites greatly enhance US defense operations; however, these key assets are vulnerable to perils such as space weather and acts of aggression. Unfortunately, it is difficult to determine the cause of anomalies from the ground. What may first appear as a routine system glitch may, in fact, be something much more serious. Providing the ability to detect—and in some cases, predict—events via multiple data sources can be critical to mission success and the safeguarding of space assets. Threat analysts must be able both to distinguish external, man-made threats, natural threats, and environmental conditions from internal, satellite bus anomalies in real time and to subsequently perform mitigating actions.

Figure 1. Data fusion and resource management architecture

To address the challenges that analysts face, AFRL researchers are using a multidisciplinary approach— one that incorporates important human systems integration (HSI) considerations into the design of space data fusion and display methods. The merger of these key research areas will enable the team to devise a system that assists human operators in detecting subtleties that may otherwise go unnoticed. In April 2006, researchers demonstrated this capability at the Center for Research Support, Schriever Air Force Base (AFB), Colorado. Five small business contractors worked together under individual Small Business Innovation Research contracts to perform the associated research. This work directly supports the Space and Missile Systems Center Space Superiority Systems Wing, as well as various activities occurring within the Defense Meteorological Satellite Program's, Technology Applications Division in support of environmental space situational awareness.

The technology demonstrated thus far involves an integrated suite of multilevel data fusion capabilities, including (1) abnormality detection and classification, (2) event tracking, (3) event relationship/situation assessment, and (4) preliminary individual visualization displays for both the operator and the analyst. The system fuses data from a variety of sources, such as satellite telemetry, space weather, orbital proximity, and radio frequency input, via the Air Force Satellite Control Network Link Processing System. To enhance its event detection algorithm effectiveness, the system converts telemetry data to "Satellite as a Sensor" abnormality reporting data via a machine learning and signature clustering technique based upon historical datasets.

Figure 2. Satellite threat analyst display that depicts orbital position and event relationships

Many researchers consider data fusion a powerful technology for satellite threat detection. Event detection algorithms assess a situation according to different kinds of data by employing a network of fusion nodes that "divide and conquer" the problem. In this approach, algorithms at each fusion level have finite responsibilities, but collectively, they accommodate many different data sources having varying levels of detail and certainty. AFRL is developing a modular threat analysis architecture that will facilitate the following:

  • Autonomous detection of threats or other abnormalities
  • Abnormality tracking
  • Tracking of associated heterogeneously related events
  • Assessment of mission impact
  • Autonomous situational response recommendations and response effects assessments against commander's intent

Intelligent use of all data sources can not only help in detecting events as they happen, but can potentially uncover an adversary's probable intent. The analyst can then apply resource management algorithms that will recommend prioritized courses of action. By autonomously detecting abnormalities, assessing the impact, and recommending a response, the data fusion tool suite will increase an analyst's situational awareness and response capability in overseeing an increasing number of space assets. Figure 1 illustrates AFRL's data fusion architecture.

The development team demonstrated three levels of data fusion at the Schriever AFB facility. Level 0 fuses the data from event reports, while level 1 tracks these reported events over a period of time. The algorithms embedded in level 2 assess both the situation and the event relationships to identify the cause of any potential abnormality.

In addition to the integrated suite's data fusion functionality, the team demonstrated the product's HSI features. Researchers established the foundation for these human-system interfaces through numerous interviews with subject-matter experts in various areas related to satellite operations, space situational awareness, and defensive counter space. Their goal was to create a common, work-centered interface rather than multiple, specialized applications.

Since many people with distinct specialties are involved in identifying and classifying a space asset attack, the team created two different human-system interfaces, one tailored for satellite/network engineers and space weather experts and a second geared towards satellite operators and threat analysts (see Figure 2). They developed the two displays according to each group's unique information needs and work habits and also designed in a certain level of user customization. The team gave each group a drill-down capability, which permits users to locate information at different levels within the system architecture. This capacity to access progressively detailed information provides historical context for data and also capitalizes on a given user's specific expertise. For example, a satellite telemetry interruption may actually stem from ground source interference rather than a satellite disruption. A threat analyst may actually have relevant knowledge of network operations—knowledge that may help him or her identify the cause even before it becomes evident to engineers. The research team is exploring future capabilities that include (1) drill down on real data for validation, (2) advanced visualization, (3) additional data sources, (4) interactive playback, and (5) enhanced assessment.

Mr. John Ianni, of the Air Force Research Laboratory's Human Effectiveness Directorate, and Mr. Paul Zetocha, of the Air Force Research Laboratory's Space Vehicles Directorate, wrote this article. For more information, visit http://www.afrl.af.mil/techconn_index.asp . Reference document HE-H-06-03.