Co-Orbital Threat Prediction and Assessment

The modern space environment is dynamic and contested, requiring constant surveillance to detect maneuvers and thwart malicious intents. Modern maneuver detection techniques rely on either increasing solution noise or estimating maneuvers by propagating multiple models. Both suffer during long data gaps or in cluttered environments, thereby yielding inconsistent results in proliferated orbital regimes. By combining Random Finite Set theory and Joint Input State Estimation, a hybrid Bernoulli filter will be built to probabilistically detect anomalies in spacecraft tracks using heterogeneous sensor data and determine if the tracks vary from the long-term mean. This will increase accuracy, reduce computational load, and allow for faster threat detection. It is unrealistic to predict all possible future behaviors of every satellite on orbit in real time so a threat model is required to probabilistically predict such outcomes. This will give a statistical likelihood of threats based on a set of parameters. During this Phase I, Tau and Texas A&M will develop this hybrid filter and design a threat posture model to accurately predict future malicious actions.