SBAS Integrity Verification Based on the Extreme Value Theory

Abstract

In many areas such as safety of life application, the integrity of the navigation system is a very important indicator to indicate the trustworthiness of the system services. According to different scenarios, system integrity requirements are also different, but are usually in the order of 10⁻⁷ or even less. It is impossible to evaluate the performance by the traditional statistics which using the frequency of occurrence to simulate the probability. The Extreme Value Theory (EVT), which enables extrapolation of the observed error distribution’s tail into the misleading information domain, regardless of the underlying error distribution. In this paper, EVT is applied to the SBAS integrity evaluation field, which can effectively solve the problem of insufficient tail samples and realize a reasonable inference of extreme abnormal events. The paper first briefly introduces the basic concept of SBAS integrity and EVT, and then uses the mixture extreme model to solve the problem of threshold choosing. Finally, the paper uses EGNOS message data provided by EMS, and wtza station observation data to evaluate its integrity performance. The result shows that the EGNOS is compliant with LPV requirements in the vicinity of wtza station.