Modeling Inertial Sensor Errors Using Autoregressive (AR) Models

Abstract

ABSTRACT: The residual random errors of the inertial sensors of a strapdown inertial navigation system (SINS) may deteriorate the overall positioning accuracy. These errors are modeled stochastically and included in the SINS error model. A first-order Gauss-Markov (GM) model with large correlation time is usually used to describe the stochastic nature of these errors. However, if the autocorrelation sequences of these random components are examined, it can be determined that a first-order GM model is not adequate to describe such stochastic behavior. This paper offers an alternative method of modeling the inertial sensor random errors as an autoregressive (AR) process. Among the different techniques that can optimally determine the AR model parameters, Burg's method was able to provide the highest accuracy when applied to navigation-grade SINS. Considering the SINS AR model for each inertial sensor, the results showed a major improvement in positioning accuracy compared with the commonly used GM models.