LLL reduction algorithm has been used as a new technique of decorrelation to GNSS ambiguity resolution for recent years.The basic idea of this method is to make the variance-covariance matrix as orthogonal as possible by virtue of integer Gram-Schmidt orthogonalization,based on this we also refer to as LLL-IGS.Although LLL-IGS can indeed be used for decorrelation,the experiments indicated that it performs worse and deteriorates in some cases,especially for real GNSS data.In this contribution,(i)A modified LLL-MIGS decorrelation algorithm is proposed by improving the sorting method and removing the error of orthogonalization.(ii)The time efficiency is introduced as a new assessment criterion to measure the performance of the decorrelation algorithm directly.