Item

Abstract

PPP-RTK, a synthesis of the well-known precise point positioning (PPP) and real-time kinematic (RTK) techniques, requires precise satellite-related products and atmospheric corrections estimated in a global navigation satellite system (GNSS) network. For generating these products, one commonly sends all network data to a computation center and conducts a centralized processing scheme. However, this centralized algorithm faces considerable challenges in computational efficiency and model formulation with expanding network coverage and increasing number of receivers. This work proposes a decentralized algorithm that generates PPP-RTK products in three steps: subnetwork processing, satellite-related product integration, and atmospheric correction update. The first step divides the whole network into several subnetworks where we can guarantee common satellite visibility at all receivers to estimate continuous satellite phase biases and impose ionosphere-weighted constraints to improve the precision of ionospheric estimates. The second step integrates subnetwork-dependent satellite clocks and biases and aligns the underlying datum. The final step updates the estimated atmospheric delays by considering the covariances between satellite-related products and atmospheric delays. For numerical evaluation, we collected one-week dual-frequency global positioning system (GPS) data from 82 stations to generate network products and conduct user positioning. The results show, on the network side, that the decentralized PPP-RTK generates products with the same precision as that of centralized PPP-RTK and reduces the computation time by 50%. On the user side, the decentralized PPP-RTK performs as well as the centralized PPP-RTK, and the time-to-first-fix and three-dimensional root-mean-square (RMS) is less than 3 epochs and 1.59 cm on average, respectively.