Abstract: As one of the key candidate technologies for 5G, the non-orthogonal multiple access technology (NOMA) employs the power domain reuse to raise the system throughput and frequency utilization effectively. Among the existing works about the NOMA system, it has been mostly assumed that the folding signal is completely synchronous. However, in the actual applications, the folding signal may not be completely synchronous due to the difference of users’ location and status. Aiming at the time-delay problem of folding signal transmission during different users, this paper proposes a new successive interference cancellation algorithm. Firstly, the superposed signals are asynchronously sampled and a series of constructional output data are obtained. And then, according to the overlap part of each adjacent symbol, a symbol is reconstructed by adopting a new successive interference cancellation strategy in the receiving terminal. It is shown from the results that the proposed method can obtain cleaner exceptional properties of the interference signal cancellation at the cost of adding complexity of signal reconstruction. Since this structure will result in the error expansion problem and affect the detection and reconstruction of all subsequent signals, it is more stringent for the requirement on the accuracy of each signal estimation and the channel estimation. Hence, this paper adopts the scalar Kalman filter algorithm to estimate the channel status information such that more accurate channel states in the time direction can be obtained. These can bring great improvement on the error detection performance of the entire signal detection such that the proposed algorithm in this paper is more applicable to the actual signal estimation.