Using Allan variance to evaluate the relative accuracy on different time scales of GNSS/INS systems

The integration of the Global Navigation Satellite System (GNSS) with an inertial navigation system (INS) can provide accurate and reliable navigation information using a data fusing algorithm such as a Kalman filter. There has been much interest in the absolute accuracy with dominant components of mid-term and long-term errors in current research works related to the GNSS/INS navigation systems, whereas few research works focus on the relative accuracy on different time scales. However, new applications of GNSS/INS integration, such as position and orientation system for mobile mapping systems and the GNSS/INS deeply-coupled receiver, require the relative accuracy on different time scales, especially that of the short-term accuracy. This paper raises the importance of the short-term accuracy of the GNSS/INS systems in certain applications. Current methods to evaluate the navigation accuracy are mainly to provide some statistical values that can represent the overall error level (e.g., RMS), but cannot show the relative accuracy. Allan variance is a method of representing root mean square (RMS) random drift error as a function of average time. An Allan variance analysis method is proposed in this paper to evaluate the relative accuracy on different time scales of GNSS/INS systems. The feasibility of the idea was verified by the road test results of different grades of GNSS/INS systems. The results show that Allan variance can give the levels of the navigation accuracy of the GNSS/INS systems on different time scales in an effective way, especially for the short-term accuracy.