Stochastic Modeling Based DGPS Estimation Algorithm

Abstract

A Kinematic Differential GPS algorithm is presented. Specifically, the accurate relative (and absolute) positioning of a team, or formation, of mobile vehicles is considered and a general navigation web concept is advanced. The measurement situation is modeled under a stochastic framework, and, rather than differencing and double differencing, as is current practice in conventional DGPS, the common errors are explicitly acknowledged and a centralized position estimation algorithm is derived. In addition, the predicted covariance of the team members' position estimation errors is obtained. Also, treating the common errors as parameters to be estimated yields estimates of the clocks' relative errors. Extensive simulations are performed to validate the Kinematic DGPS algorithm. The results are compared to conventional DGPS scenarios, where the reference stations are stationary or there is only one available observation epoch. The positioning accuracy improvements achieved are gauged against the performance of conventional DGPS.