Abstract:

For the multi-system GNSS RTK feature,the integer ambiguity is introduced in this paper.However,due to the increase of visible satellite,the dimension of the state vector will increases sharply,which leads to a large number of multi-dimensional matrix multiplication.And the covariance calculation and storage capacity account for 70% of the total Kalman filtering process time.In order to improve the efficiency of the algorithm,an improved Kalman method for sparse state transition matrices is proposed.This paper mainly discusses the solution of covariance matrix and uses matrix block units with symmetry to reduce the multiplication times to less than 10% of the general algorithm.The experimental result is that the improved method of CPU processor takes less than 10% of the general algorithm.This improved method achieves high efficiency,which also has certain reference value for other filtering algorithms.

Key words: multi-system, GNSS RTK, covariance matrix, iteration, matrix multiplication