Abstract: A modern cross spectrum estimation method is proposed, which is the SVD algorithm based on cross-spectrum of the Auto-Regressive (AR) model for parameter estimation. Compared to previously proposed Levinson algorithm based on cross-spectrum of the AR model, the new method of this study can effectively overcome the influence of the estimation error of the cross correlation function r_xy(m). Simulation results show that the spectrum estimation accuracy of the proposed method is much better. A simulation platform in the measurement noise background is established. The anti-jamming performance of the Transform Domain Communication System (TDCS) based on the proposed method is studied. The results show that the TDCS can effectively suppress the measurement noise, greatly reduce the error rate under different disturbances, and improve the capacity of antidisturbance.

Key words: communication technology, transform domain communication system, autoregressive model, error rate