A Parallel Processing Strategy of Large GNSS Data Based on Precise Point Positioning Model

摘要

With continuous increasing of the data scale of GNSS observations network, the computing pressure of data processing is growing. The undifferenced precise point positioning (PPP) model is one of the main strategies of GNSS network data processing. With the increasing of stations' scale, the processing time of PPP pattern also increases linearly, the traditional serial processing pattern need to consume a large amount of computing time. As the PPP model is not related, this model has good characteristics of parallel processing between stations. This paper established a distributed parallel processing strategy based on the PPP model, which can not only improve the efficiency of data processing, but also enhance the efficiency of hardware performance. However, due to the high concurrency of data access and processing, the parallel programming is faced with great challenges which can cause immeasurable results. In this paper, by analyzing the flow characteristics of the PPP method, a parallel GNSS data process model at multi-core and multi node level was set up, and a lightweight parallel programming model was adopted to realize the parallel model. Through a large number of data tests and experiments, high efficiency of parallel processing of GNSS data based on the PPP model was achieved. The experiment shows that, under the environment of four multi-core nodes, the parallel processing is at least six times faster than the traditional serial processing.