GNSS/IRS Hybridization: Fault Detection and Isolation of More than One Range Failure

GPS/IRS hybridization is a good candidate to fulfill demanding civil aviation requirements. When the integrity of the GPS measurements is ensured they may be used to calibrate inertial position and improve accuracy. This can be done in a tightly coupled manner by means of a Kalman filter. Calibrated IRS can ensure coasting while maintaining good short term accuracy and helps detect large GPS failures. The hybridized system must also be able to detect slowly growing errors on GPS measurements that may affect inertial calibration. These errors may affect one or several GPS measurements. Fault detection and exclusion capacity of Receiver Autonomous Integrity Monitoring is limited for GPS, as it is designed for one satellite failure only. Therefore other solutions have been proposed. Among these are the extrapolation method and the solution separation method that are also designed to detect one satellite failure at a time. The study reported in this paper presents first results obtained with an extension of Mats Brenner’s separation algorithm for fault detection. That extension concerns detection and isolation of more than one failure during operations ranging from en-route to NPA, with the objective of satisfying the availability requirement. The paper starts with a brief review of the classical Solution Separation method for a single satellite failure detection and isolation. In this section the model of the tightly coupled GPS/IRS system involved in simulations is also presented. The third section is dedicated to theoretical description of the one failure Solution separation fault detection and exclusion algorithm. The next section considers the behavior of the algorithm when two simultaneous range failures occur. In particular the impact of two faulty measurements on the test metrics in the former fault detection and exclusion algorithm is examined. These simulation results are used to discuss the contribution of a two range failure detection algorithm. Afterwards, a two simultaneous range failure detection and exclusion algorithm based on an extension of the nominal Solution Separation method is introduced. Definitions of performance parameters (Horizontal Protection Level, Horizontal Exclusion Level, and Missed Detection Probability), when several pseudo range failures (ramp, bias, etc.) occur simultaneously, are discussed. Finally initial simulation results are presented. Presented at ION GPS 2002, Portland 1