Abstract: In the field of non-cooperative communication, the traditional OFDM time parameter estimation algorithm based on cyclic autocorrelation analyses the OFDM signals without pilots, which is not suitable for the parameter estimation of the OFDM signals with pilots in the actual application scene. In this paper, the cyclic autocorrelation spectrum characteristics of OFDM signals with different pilot patterns are analyzed, and the OFDM signals parameter estimation algorithm is improved for changes of the spectral feature. The improved algorithm is applicable to both OFDM signals without pilots and OFDM signals with different pilot patterns. Simulation results show that, for OFDM signals without pilots, the improved algorithm can reduce the time parameter estimation error by one order of magnitude over the traditional algorithm. At the same time, the improved algorithm effectively avoids the influence of the two-order periodic spectral peaks introduced by pilots to the estimation of OFDM time parameters. It can accurately estimate the time parameters of the OFDM signals with pilots, and it has good robustness and application value.