Advanced Receiver Autonomous Integrity Monitoring in Tightly Integrated GNSS/Inertial Systems

Abstract

This paper deals with the tight integration of Inertial Measurement Units (IMU), Global Navigation Satellite Systems (GNSS) and the concept of Advanced Receiver Autonomous Integrity Monitoring (ARAIM). While the integration of an IMU and GNSS to an integrated GNSS/Inertial system is well known and widespread, the use of the ARAIM concept in inertial systems is a new and promising approach. In safety critical applications such as aviation, GNSS receivers as well as integrated GNSS/Inertial systems have to be equipped with a Fault Detection and Exclusion (FDE) function. Single frequency L1 GPS receivers with Receiver Autonomous Integrity Monitoring (RAIM) were the answer for decades. The second generation of GNSS offers more satellite systems and more frequencies for navigation. The visibility and accuracy of Multi-Frequency and Multi- Constellation (MFMC) receivers are significantly improved. ARAIM transfers these improvements into aviation. MFMC receivers with ARAIM can provide protection levels for challenging Alert Limits, for instance VAL = 35 m, with reasonable availability. Therefore, ARAIM has the potential for LPV-200 (Localizer performance with vertical guidance, decision height 200 feet) approaches. By including an IMU, it is possible to increase this potential. The added value of using an IMU and the difficulties of integration are hardly mentioned in literature. ARAIM ensures integrity by comparing the GNSS position solution with all satellites in view to solutions of subsets (fault-tolerant solutions) that exclude certain satellites. A transfer of this concept into a tightly integrated GNSS/Inertial system seems straightforward – replace the GNSS position solutions and subsets by integrated GNSS/Inertial position and subsets. On the other hand, ARAIM needs to evaluate hundreds of subsets, which creates a considerable computational load, especially in the case of GNSS/Inertial integration. A challenge for GNSS/Inertial designs is the ARAIM specific ranging model, which includes ranging bias, accuracy and integrity. Carrier smoothed ranges are used in ARAIM. These smoothed signals contradict an optimal GNSS/Inertial integration. In addition, the integration design has to consider time correlations of ranging signals, neglectable for ARAIM. In this paper, we address the mentioned design issues of a tight GNSS/Inertial integration, which uses the concept of ARAIM. We also describe our simulation procedure to determine availability. The availability serves as performance measure for the integration designs. Benefits and effects of reducing the number of subsets, interpretation and implementation of ranging model, correlation time constants, as well as different IMU classes are investigated by means of simulation results.