返回Google圖書搜尋

Analytical Estimation of Carrier Multipath Bias on GPS Position Measurements

出版	National Institute of Standards and Technology, 1994
URL	http://books.google.com.hk/books?id=oIuxoQEACAAJ&hl=&source=gbs_api

註釋Multipath is one of the factors degrading the accuracy of position measurements obtained with the Global Positioning System (GPS). We investigate the effects of multipath on the carrier phase measurement and the resulting bias on relative GPS positions for observation times longer than several hours. A short-range GPS network was surveyed with day-long observation sessions. The data display obvious multipath signatures and the position results indicate the presence of site-specific errors at the level of several millimeters. This led us to suspect multipath bias and motivated the quantitative estimation presented here. We first model the phase error due to multipath from a single plane reflector in terms of satellite reflector geometry. The effect of this error on position is then derived by performing a simplified least squares adjustment under the assumption of uniform satellite distribution. After several steps of approximations and manipulations we arrive at simple expressions for the bias in terms of general receiver-reflector geometry and a few other variables. The results are generalized to dual frequency observations that are commonly used for high accuracy observations. The model is then employed to estimate upper limits of the multipath bias for general receiver environments. We consider bias due to multipath from the flat ground, from nearby objects, and from a tilted ground, obtaining formulae for each situation that depend only on reflector characteristics and geometry. Finally, these results are applied to the GPS experiment. Limitations of the applicability due to model assumptions and unmodelled effects are discussed. The two stations thought to be most affected by multipath in the experiment are examined. We find that the multipath phase error as inferred from the observations can give rise only to vertical biases smaller than 2mm and horizontal biases smaller than 1mm. It is thus concluded that bias due to multipath-induced carrier phase error is effectively reduced by averaging if observation intervals are at least several hours long and son basic precautions are taken regarding the receiver environment.