Efficient, Covariance-Constrained Sensor Fusion Through Observation Decimation

Spacecraft operate under limited on-board resources resulting from size, weight and power (SWap) requirements. These on-board limitations include limited data processing capability. As spacecraft operations complexity grows so does the reliance in system to accurately provide the internal state of these systems from sensor data. Estimators used to reconstruct the internal state of a system rely in sensor data assimilation to produce accurate state reconstruction. We explore estimators for linear systems that reduce the amount of data they need to assimilate while keeping a desired level of performance in the reconstruction of the internal state of a system. As a measure of estimator performance, we employ the state covariance. We use the discrete algebraic Riccati equation to predict the steady state covariance of a Kalman Filter implementation of a linear system. To perform data assimilation reduction, we implement a measurement update decimation that depends on the desired level of estimator performance. To test this approach, we implement a Kalman Filter to an arbitrarily large linear system that implements the maximum allowable measurement update decimation while maintaining the estimator covariance within a maximum desired level.