ABSTRACT

A modified algorithm combining equivalence theory with MPI technology to improve Global Navigation Satellite System data processing efficiency is presented. The reservation of covariance matrix, elimination of unknowns and parallelism of the equivalence algorithm are derived. Experimental results show that the algorithm can improve the efficiency of processing 100 IGS stations by 56% with negligible precision loss. Moreover, the proposed approach has higher efficiency than the leastsquares and Gauss-Jordan parallel method and this advantage is more obvious as computational scale increases. Further experiments on a clustershow the more nodes used, the higher the efficiency it can achieve.

摘要

提出了一种将等价理论与MPI技术相结合的改进算法，以提高全球导航卫星系统的数据处理效率。推导出协方差矩阵的保留、未知数的消除和等价算法的并行性。实验结果表明，该算法在精度损失可忽略不计的情况下，将处理100个IGS站的效率提高了56%。此外，所提出的方法比最小二乘法和高斯-乔丹并行方法具有更高的效率，并且随着计算规模的增加，这种优势更加明显。在集群上的进一步实验表明，使用的节点越多，它可以实现的效率就越高。