Convergence of Block Decorrelation Method for the Integer Ambiguity Fix

Because of the integer-valued nature of carrier phase ambiguities, it is essential to fix the float estimates into integer values in order for high precision DGPS positioning. A decorrelation process is necessary to solve the problem since double-differenced ambiguities are highly correlated in general. In this paper, Block Decorrelation Method (BDM) is presented and tested for its convergence. BDM divides the variance-covariance matrix into four blocks and decorrelates them simultaneously. A number of randomly selected examples show that BDM is comparable to the existing decorrelation algorithm, however its speed of convergence is relatively faster due to the computations performed on small blocks.