Focused Real-time Systems Monitoring Based on Multiple Anomaly Models

In real-time monitoring situations, more information is not necessarily better. When faced with complex emergency situations, operators can experience information overload and a compromising of their ability to react quickly and correctly. We describe an approach to focusing operator attention in real-time systems monitoring based on a set of empirical and model-based measures for detecting different kinds of anomalies and for determining the relative importance of sensor data . This approach has been evaluated on data from the life support system testbed of Space Station Freedom .