An Experimental Study on Adaptive Sequential U-D Filtering and Propagation of SINS Errors during GNSS Outage

Strapdown inertial navigation system (SINS) is used as a primary navigation information source on board an aircraft and is expected to provide high accuracy navigation solution. Often, the pure-inertial navigation solution is blended with global navigation satellite system (GNSS) data through optimal filtering to provide bounded and accurate navigation information. Several adaptive Kalman filtering (AKF) algorithms published earlier have considered either the modelling or estimation of measurement error covariance matrix R_k and process covariance matrix Q_k. The simultaneous estimation of both R_k and Q_k is limited in their performance due to instability for long endurance high accuracy navigation applications. The measurement noise covariance matrix R_k under all practical conditions is influenced by external factors. In this manuscript, the adaptive estimation of R_k has been explored along with the accurate computation of Q_k. Further, an attempt has been made to propagate the error state covariance during GNSS outage with an accurate modeling of system matrix and corresponding Q_k matrix computations. The sequential U-D filtering approach is explored to handle the ill-conditioning of P_k. The effect of propagation of P_k is judged through the quantification of pure-inertial navigation drift rate under GNSS outage conditions which is further decided by the quality of estimation of sensor biases. The effectiveness of these estimated sensor biases along with adaptive estimation of R_k and computation of Q_k is demonstrated through aircraft flight testing. Finally, various AKF algorithms are validated along with the propagpropagation studies and conclusions are drawn for practical use.