Advanced RAIM Algorithms: First Results
Résumé
Receiver Autonomous Integrity Monitoring (RAIM) with detection/ exclusion functions is currently a major technique for the GNSS in many safety-critical civil aviation applications. Although RAIM algorithms are not directly standardized, most existing implementations are based (or equivalent to) a snapshot technique based on the least squares (LS) algorithm. The principal limitation of these conventional RAIM algorithms is the availability of detection and especially exclusion functions. This problem becomes crucially important when the RAIM has to be used to provide integrity for the vertical positioning information (for example to support APV I & II approaches). Therefore DGAC/STNA has sponsored UTT to conduct a study to investigate the potential benefits of more advanced RAIM algorithms to support APV operations. First, a statistical model of the position failure detection and exclusion based on RTCA high level definitions is defined. This model provides us with a mathematically formalized description of different events, which is important for further RAIM’s algorithm design, optimization and comparison. A formalized definition of a horizontal/vertical error that must be considered as a positioning failure has been proposed. This definition is used to compute for each epoch and for each satellite channel a pseudorange bias that leads to a positioning (horizontal/vertical) failure. A very special attention has been paid in this study to estimate the impact of the pseudorange autocorrelation function, due to the “tropo" and “iono" refraction, on the statistical performance (and availability) of RAIM schemes. No result is really available in the literature on this problem, which is crucially important for new types of high accuracy channels (like GPS L1/L5 or Galileo E1/E5). Next, we briefly introduce and describe two new RAIM detection/exclusion algorithms: snapshot RAIM algorithm based on the constrained Generalized Likelihood Ratio Test (GLRT); sequential RAIM algorithm based on the constrained GLRT. We also use the principle of pre-filtering in order to improve the signal-tonoise ratio of the faults. Finally, we analyze and compare the performance of the conventional LS-based RAIM detection/ exclusion solution with the proposed snapshot and sequential constrained GLRT-based algorithms with and without pre-filtering.