Threat Assessment and Validation Toolset

We propose a toolset and architecture that will not only provide assessment and mitigation of threats, but will also provide a simulation environment to allow generation of realistic data sets and metrics for testing and comparison of different threat detection and assessment algorithms. Our approach differs from previous attempts because it is based on ensembles of decision trees that are tractable and utilizes actual features in the telemetry that can be characterized quantitatively (as opposed to neural net approaches). Although the focus on this Phase I will be to produce the threat assessment, simulation environment, and validation performance metrics, the algorithms will be designed to operate within the larger framework of the Joint Directors of Laboratories (JDL) model for data fusion. One of the outputs of the simulation environment will be a library of anomalous data that can be used to evaluate different detection techniques and quantitatively assess their ability to detect and ultimately distinguish between technical, environmental, and actual threats to the spacecraft. The simulation environment will also allow “what if” scenarios through deterministic and non-deterministic adjustments of the input variables.