Improvement of Processing Time of GNSS Data for Point Positioning Through Parallel Computing

M. Khoshmanesh and A. Abouhamzeh

Abstract:	In today’s modern world, the matter of positioning is one of the most significant ones where high interests are drawn. Although GNSS have initiated a method in order to estimate the coordinates, the matter of processing raw data of these systems to obtain reliable outputs has been remained as a baffling problem in scientific societies. In order to answer this demand, the great potential of multi-core processers can be used in this growing domain. In this paper, the specifications that distinguish the GNSS data from other sorts of data in the potential of multi-core processing point of view are discussed. As an illustration of the significant effect of multi-core processing on GNSS data, a case study on absolute point positioning has been done. The algorithm was written in two forms which one is designed for the efficient usage of all the CPU’s cores and the other one can use only one core regardless of the number of CPU cores available. The programming environment chosen for this study is MATLAB and the multi-core progress has been done using the MATLAB parallel processing toolbox. On the other hand, due to the number of programs running at the same time and also the effect of operating system itself, the processing time was measured 10 times and the provided results are the average of all measured times. Furthermore, an analyzing process has been applied which include identifying inhibitors to parallelism and a cost weighting on whether or not the parallelism would actually improve performance. For this purpose, codes of the program of point positioning are classified based on the processing instructions and the potential of parallelism has been investigated in these groups. The effect of parallelism on each of these code groups has a completely difference behavior compared to other ones as the consumed time changes between 75 percent decrease to more than 1500 percent increment. Considering the results of this analysis, the serial code of the point positioning problem has been revised aiming to achieve the optimum results in processing time point of view. Finally, the results of parallelism on the point positioning problem show more than 30% improvement when only 2 independent physical cores are utilized. As the number of independent physical cores of the processor increases to 4, the running time improves by 50% and 27% compared to serial code and 2 independent cores mode respectively.
Published in:	Proceedings of the 25th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS 2012) September 17 - 21, 2012 Nashville Convention Center, Nashville, Tennessee Nashville, TN
Pages:	2239 - 2248
Cite this article:	Khoshmanesh, M., Abouhamzeh, A., "Improvement of Processing Time of GNSS Data for Point Positioning Through Parallel Computing," Proceedings of the 25th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS 2012), Nashville, TN, September 2012, pp. 2239-2248.
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In