Global Navigation Satellite System (GNSS) data from reference station, networks deployed globally can facilitate positioning, navigation and timing applications. To enable precise positioning for dual-frequency users, several representative methods relying on GNSS reference networks have been developed, such as Network Real Time Kinematic (NRTK), Precise Point Positioning (PPP) and PPP-RTK. The state-of-the-art PPP-RTK integrates the advantages of customary NRTK and PPP, and has become an important topic in current research. In this contribution, a network processing method is proposed to achieve single-frequency PPP-RTK. The elementary procedures are as follows: 1) A Kalman-filter-based customary PPP is implemented station by station, with known geometric ranges and satellite clocks fixed. The estimable unknowns consist of, among others, the ionospheric delays and the float-valued carrier-phase ambiguities. 2) After measurement-update, the filtered PPP ambiguities of all stations are incorporated and reformulated into three sets of new parameters, namely, double-difference (DD) ambiguities, receiver and satellite carrier-phase biases. 3). The reformulated DD ambiguities are resolved into integers, and then the satellite carrier-phase biases as well as those filtered ionospheric delays are further updated. On the user side, by applying the satellite phase biases and (interpolated) ionospheric delays, the integer ambiguity resolution enabled single-frequency PPP-RTK is fulfilled. Numerical tests using daily GPS data collected by an Australian Continuous Operating Reference System (CORS) network and a single-frequency (u-blox) rover receiver show that success rate of CORS network ambiguity resolution is as high as 98. 89%. In addition, the stability of estimated satellite carrier-phase biases is better than 0. 2 cycles over every continuous satellite arc. By confronting the ionospheric delays interpolated from the CORS with that determined from a dual-frequency receiver co-located with the rover receiver, interpolation error of 10 cm has been verified. Re-initialization Of Kalman-filter-based single-frequency static/kinematic PPP-RTK is attempted at every epoch, and the resulting time-to-first-fix values, as a measure of the time required for integer ambiguity resolution, are never more than 10 min. With the aid of resolved integer ambiguities, the RMS of single-frequency kinematic PPP-RTK positioning errors becomes as good as 5 cm for horizontal component and 10 cm for vertical component. Before ambiguity resolution, these RMS values vary from 28 to 53 cm. Although it is developed with the goal of enabling single-frequency PPP-RTK, the network processing method proposed does not lose its ability to attain dual-frequency PPP-RTK capability. More importantly, this method also reserves simplicity as well as flexibility in multi-frequency, multi-GNSS applications.
