Research On Software GPS Receiver Positioning Algorithm

At present, Global Positioning System (GPS) is the most advanced satellite navigation system. For its characteristics of providing global, continuous, real-time and high accuracy 3-D position and velocity as well as time, GPS has been widely applied in many fields, and relative products emerged in endlessly that involve a lot of fields such as vehicle navigation, geographic position and so on.However, the positioning accuracy of the GPS, especially the low-price GPS receiver, cannot always be satisfied and the positioning error may be about 10 or 15 meters. This is because during the positioning process many errors, especially the measurement and satellite position errors, cannot be deleted. Traditional methods can hardly minimize their influence. Filtering algorithm provides an important method to reduce these errors and improve the GPS positioning precision.One of the generally used optimal filtering algorithms is Kalman filtering. However, the standard Kalman filter can only be used to solve the linear problem, not solve nonlinear problems. While the extended Kalman filter which linearizes the model affected the performance of the system in a certain extent. So, recently, the unscented Kalman filter which could use directly on non-linear problem is gradually becoming a useful method of GPS positioning.This paper firstly reviews the development of GPS history, and then discusses briefly the basic structure of GPS receivers and positioning principles. Then GPS positioning model is presented, which takes advantage of high-accuracy UKF as the positioning algorithm in the hope of getting more accurate position information.However, Kalman filter is an increased memory filter, when the principle of dynamic system model is precisely known without computation errors. More accurate estimation will be provided gradually. Because the norm of estimation error covariance is decreased progressively, the effect of new observations data for amending state estimation will be weakened. But in fact, the changes of system dynamic model are difficult to fully know in advance. On the one hand the parameters of systems is changing, on the other hand, in the recursive process, these changed parameters have been computed as a precise value. As a result, the norm of the error covariance matrix decreased over time, the actual error covariance is increasing, and eventually lead to filter divergence. Attenuation factor have been added in the kalman filter, which named attenuation memory Kalman filter (MAUKF). MAUKF makes the weight of new measurements increased, while the weight of the old measurements gradually decreased. By experimental verification, its positioning accuracy is higher than Unscented Kalman filter.The results of this work can be used to optimize software GPS receivers'positioning performance as well as to improve positioning methods of high-quality receivers.