In sparse code division multiple access (SCMA) systems, because multiple user data are superimposed on the same subcarrier and the channel conditions of different users are different, accurate channel estimation cannot be performed directly. To solve this problem, a channel estimation method for SCMA systems based on orthogonal superposition is proposed. This method uses the sparsity of SCMA codebook to design a new pilot structure with multi-user timeslot multiplexing, which can ensure the orthogonal subcarrier resources and reduce the timeslot overhead of pilot resources after the same timeslot is superimposed on the pilot data of different users. On this basis, a pilot structure of local orthogonal superposition is proposed, and the channel fading coefficients of adjacent subcarriers are estimated by using the correlation between subcarriers to further reduce the time slot overhead of pilot resources. The simulation results show that the channel estimation MSE performance of orthogonal superposition method is improved by 3 dB and the local orthogonal superposition performance is further improved by 1.7 dB compared with the traditional method under the same total time cost.