J4 ›› 2009, Vol. 36 ›› Issue (6): 1010-1014+1020.

• Original Articles • Previous Articles     Next Articles

High accuracy complex sinusoid frequency estimator in low SNR environment

LI Xiao-jie;XU Lu-ping;ZHOU Xue-song   

  1. (School of Electronic Engineering, Xidian Univ., Xi'an  710071, China)
  • Received:2008-09-26 Online:2009-12-20 Published:2010-01-20
  • Contact: LI Xiao-jie E-mail:xjli_xa@126.com

Abstract:

By using the information on both amplitude and phase, the frequency estimation algorithm MAP/ML can achieve high accuracy frequency estimation when the SNR is high. But the performance will be greatly degraded at a low SNR. Based on analyzing the reason, a new maximum likelihood (ML) estimation method is presented, which can confirm the coherent integration length and the unwound sequence combined with FFT. Using the coarse frequency estimation value for frequency shift, signal's frequency and the variation of sequence's phase are limited to a certain sphere, so that the Cramer-Rao lower bound(CRLB) limits with various coherent integration lengths are derived, and the optimal coherent integration length of data subsection is determined, which contributes to attaining the optimal balance between the estimation accuracy and the SNR gain. In addition, an estimation sequence selecting method based on the phase variance of the phase-shifted sequence is proposed, which avoids the phase unwraping process. Finally, the Monte Carlo simulation result shows that the performance of the estimator can approach the CRLB very closely.

Key words: frequency estimation, coherent, Cramer-Rao lower bound(CRLB), phase unwraping

CLC Number: 

  • TN971