Multipath is a major source of error in high precision Global Navigation Satellite System (GNSS) static and kinematic differential positioning in urban environments. This paper describes a unique approach to mitigate strong multipath error: a new multipath mitigation technique ensuring that antenna motion for a rover-moving platform is maintained in the case that the platform is moving slowly or is stopped. It is known that standard GNSS receivers are vulnerable to multipath interference when the rover antenna is static. This is because stable and strong multipath signals can be easily received when the antenna is not moving, as the carrier phase relationship between the direct signal and the reflected signal changes slowly. Conversely, when a vehicle is moving, the received carrier phase relationship between the direct signal and the reflected signal changes rapidly, meaning that the strong reflected signal will be averaged or disappear. We attempt to use this characteristic to mitigate strong multipath errors. Three experiments were conducted to evaluate the proposed technique. The first test results illustrate the case of receiving strong specular reflection in a static condition. The proposed technique of maintaining antenna motion can reduce multipath errors from over 15 m to 1-2 m. In the second test, results represent the case of multipath mitigation in a car by comparing two closely set antennas: one is where the antenna is fixed on the roof of the car; the other is where the antenna is intentionally shaken manually while the car is stopping. The latter case can reduce significant multipath errors that occur while a vehicle is stopping at an intersection traffic signal. Finally, we propose a new approach to mitigate strong multipath errors corresponding to speed without maintaining antenna motion. We modify the positioning algorithm according to the speed. Test results confirm that the proposed method can reduce multipath errors effectively, especially in the case where the platform is moving slowly. The standard deviation of total absolute horizontal errors are reduced approximately 20%.
