Abstract: Aiming at the estimated accuracy problem of time of arrival (TOA) of the pulse signal processed by channelization under low signal-to-noise ratio (SNR), an efficient method for extracting the time of arrival of the pulse signal is proposed. Firstly, the received data is channelized to extract the pulse signal. Secondly, the pulse signal is preprocessed by correlation, and then the signal envelope is extracted to estimate the coarse time of arrival. Based on the above operation, set cascaded sliding window to calculate energy at the signal frequency point according to the original sampling data, and the time of arrival of pulse is refined by the energy change at the signal frequency point. Compared with traditional methods, proposed method has higher estimation accuracy at low SNR and less computation. Validity of this method is proved by simulation results.