In this paper, A low complexity decoding scheme of raptor-like low-density parity-check code (LDPC) in sufficient SNR scenarios such as in an over-engineered network or core layer of layered-division multiplexing (LDM), is proposed. This scheme is meaningful to emerging power-limited devices, such as IoT and handheld devices. Based on the proof of boundedness and convergence properties of raptor-like LDPC code, the proposed decoding scheme selectively skips some check node operations during iterative decoding. Furthermore, we propose a corresponding algorithm to choose a proper set of skipped check nodes, which ensures an acceptable performance with much lower complexity. Simulation results show that the proposed scheme is good enough to meet the performance requirements, while yielding attractive power benefits over conventional scheme. The decoding complexity can be reduced by up to 80% in the best case while the SNR is sufficient.