The pre developed global navigation satellite system (GNSS) dual frequency sum diff joint tracking algorithm has the shortcoming of poor tracking robustness and tracking precision of dualfrequency signals in the case of inconsistent carrier-to-noise ratio (CNR) of dual frequency signals. An optimized and improved the original sumdiff joint tracking algorithm method is proposed and the main idea is to optimize the original dualfrequency sumdiff joint tracking loop structure for different CNR of the common satellite in practical applications. The weighted value of the GNSS dualfrequency signal in the dualfrequency sumdiff joint tracking loop is calculated based on the CNR weighting model, and the carrier joint tracking of the dualfrequency signal is achieved by using the weighted value. The algorithm simulation results show that the improved GNSS dualfrequency sumdiff joint tracking algorithm based on CNR model weighting is more in line with the practical application, and can improve the tracking robustness and precision of the original algorithm in the case of inconsistent CNR of dual frequency signals.