Layered decoding algorithm layers the check matrix and updates the messages by layer, which accelerating the decoding convergence rate. However, since message updates in each layer is based on the upper layer, it is unavoidable to introduce the irrelevant information of the upper level, which reduces the proportion of effective details of the current layer. Aiming at the problem, this paper presents a layered-sorted min-sum ascend (LS-MS-A) decoding algorithm that introduces an ascending layer-weight scheduling strategy. It first updates the layer with smaller weight and reduces the interference of irrelevant information from the upper layer to the effective information of the current layer, thus, speeding up the decoding convergence rate. According to the simulation results, under the premise of ensuring the system reliability, compared with the layered minimal sum decoding algorithm, when the code length is 256 and the code rate is 0.5, LS-MS-A can improve the decoding convergence speed by about 9%. When up to 512 and the code rate is 0.75, the decoding convergence speed can be improved by about 15%.