基于UKF的车辆GPS/蜂窝网无缝定位算法

Vehicle GPS/cellular network seamless positioning algorithm based on UKF

当车辆运行在存在遮挡和多路径的复杂环境下时，车载GPS接收机由于无法接收到足够数量的GPS定位卫星将导致车辆无法准确定位。GPS/蜂窝网无缝定位技术将蜂窝基站视作伪卫星，将蜂窝基站观测值与 GPS 观测值结合进行融合解算有效解决了该问题。针对目前融合解算算法所采用的 EKF解算算法存在高阶项截断误差和 Jacobian 矩阵难计算的缺点，本文将不敏卡尔曼滤波(UKF)引入到基于GPS和蜂窝网的无缝定位算法中。UKF算法是一种基于UT变换的非线性的滤波方法，避免了因进行强行线性化带来的误差影响。通过在高斯信道、静止和运动场景下，对两种混合定位算法性能进行对比分析，实验结果表明：将UKF算法用于GPS和蜂窝网络的混合定位中时，在不同的定位环境下，其定位均方根误差（RMSE）相较EKF算法降低了20%~70%。

When a vehicle runs in a complex environment with occlusions and multi-paths, the GPS receiver can not receive sufficient GPS satellites which will lead to inaccuracy in locating the vehicle. This problem can be solved by GPS/Cellular network seamless positioning technology which takes the cellular base station as a pseudo satellite and joints the observations of cellular base station and GPS together. The EKF algorithm in the present fusion decoding algorithm has higher-order term truncation error and hard-calculated Jacobian matrix. Aiming at these shortcomings, a UKF is introduced into the GPS/cellular network seamless positioning algorithm. The UKF algorithm, mainly based on UT transform, is a kind of nonlinear filtering method and can avoid the errors caused by forced linearization. Finally, the performances of two positioning algorithms are compared and analyzed in Gaussian channel, stationary and motion scene, respectively. The results indicate that, by introducing the UKF algorithm into the GPS/cellular networks hybrid location under different positioning environments, its positioning error (RMSE) has reduced by 20%-70% compared with that of EKF algorithm. ©, 2015, Editorial Department of Journal of Chinese Inertial Technology. All right reserved.