A combined algorithm of improving INS error modeling and sensor measurements for accurate INS/GPS navigation

Although the integrated system of a differential global positioning system (DGPS) and an inertial navigation system (INS) had been widely used in many geodetic navigation applications, it has sometimes a major limitation. This limitation is associated with the frequent occurrence of DGPS outages caused by GPS signal blockages in certain situations (urban areas, high trees, tunnels, etc.). In the standard mechanization of INS/DGPS navigation, the DGPS is used for positioning while the INS is used for attitude determination. In case of GPS signal blockages, positioning is provided using the INS instead of the GPS until satellite signals are obtained again with sufficient accuracy. Since the INS has a very short-time accuracy, the accuracy of the provided INS navigation parameters during these periods decreases with time. However, the obtained accuracy in these cases is totally dependent on the INS error model and on the quality of the INS sensor data. Therefore, enhanced navigation parameters could be obtained during DGPS outages if better inertial error models are implemented and better quality inertial measurements are used. In this paper, it will be shown that better INS error models are obtained using autoregressive processes for modeling inertial sensor errors instead of Gauss–Markov processes that are implemented in most of the current inertial systems and, on the other hand, that the quality of inertial data is improved using wavelet multi-resolution techniques. The above two methods are discussed and then a combined algorithm of both techniques is applied. The performance of each method as well as of the combined algorithm is analyzed using land-vehicle INS/DGPS data with induced DGPS outage periods. In addition to the considerable navigation accuracy improvement obtained from each single method, the results showed that the combined algorithm is better than both methods by more than 30%.