Abstract:There exists severe deterioration of phase noise in the ultra wideband application scenario of terahertz communication, since terahertz carriers are achieved through multiple frequency doubling. Based on the multipole/zero phase noise model at terahertz frequency, a fully parallelized phase estimation algorithm is proposed on the basis of traditional blind phase estimation algorithms. This algorithm inserts pilots into each parallel data and uses pilot phase noise information as the initial phase to perform phase extension and rotation phase detection on the parallel data. Referring to the traditional blind phase estimation algorithm, the decision selection approach for the optimal phase estimation value is adopted. At the same time, the previous and current time pilots are employed to estimate the phase of the current parallel data from the front and back directions. The two estimated phases are weighted and summed based on the distance between the front and back pilots to obtain the optimal phase estimation information. After simulation verification, the residual phase noise is reduced by 10 dBc/Hz and 25 dBc/Hz at 1 MHz and 10 MHz, respectively, through this algorithm.