Detecting Threats To PWSA Constellation Using AI and Satellite Telemetry Data

Large, proliferated space networks are here to stay.  Operators of such networks need tools to help them assess the status of individual satellites as well as the state of the constellations.  DF&NN proposes to adapt existing, self-sustaining AI/ML autoencoder style neural networks (NNs) to each satellite.  A custom NN application trained on normal telemetry for that satellite will evaluate data and determine if an abnormal event is in progress.  The events detected on each PWSA satellite not only provide real-time insight for operators, but can also be archived and autonomously analyzed.  The correlations include, but are not limited to ground station proximity, space weather events, and known adversary systems proximity.DF&NN also proposes to aggregate the state of the PWSA constellation from the status of individual satellites as well as provide the status of the PWSA constellation to provide various space support missions over selected theaters of operations or geographic regions.