Abstract:Classic Kalman filtering requires noise to be Gaussian white noise. However, the observation error and the state prediction error in GNSS kinematic positioning are colored noise. This paper establishes the colored noise model by using past observation residuals and state residuals in order to weaken the effects of colored noise on kinematic navigation solutions. Quad-constellations GNSS receiver measurements are used for a kinematic navigation experiment, and the results show that the algorithm can effectively improve positioning accuracy, as compared with the classic Kalman filtering algorithm with no consideration of the colored noise. The improvement rate of three-dimensional position accuracy is over 9%.
SUN Qingfeng,CAI Changsheng,CUI Xianqiang et al. A Filtering Algorithm for Quad-Constellation GNSS Kinematic Navigation with Consideration of Colored Noise[J]. jgg, 2019, 39(6): 625-628.