Abstract: To solve the problem of carrier frequency synchronization in HF burst-mode communication, a Non Data Aided（NDA） algorithm based on particle filter is proposed. The Maximum A Posterior（MAP） estimation of carrier frequency offset is made based on Bayesian formulation. And the minimum mean square error（MMSE） estimation of carrier phase and modulated symbols are obtained by particle filter algorithm. The particle weights are recursively computed using Sequential Importance Sampling（SIS）. Systematic resampling is employed for the reduction of particle weights degeneration. The carrier recovery performance of the algorithm for the channel with Gaussian white noise is simulated. The results show that the algorithm has short carrier recovery time and wide carrier recovery range. A HF communication simulation model is built based on Watterson model. Signals are simulated through this model under different channels, Signal-to-Noise Ratio（SNR） and frequency offset conditions. The result shows good performance of symbol detection even for large frequency offset.

Key words: HF burst-mode communication, Non Data Aided, particle filter

摘要： 针对传统短波突发通信中载波恢复及符号检测问题，提出了一种基于粒子滤波的无辅助数据的算法。首先利用贝叶斯准则实现收发频率差的最大后验估计，然后通过粒子滤波算法实现载波相位和调制符号的最小均方误差估计。算法中使用了序列重要性采样技术实现粒子权值的迭代更新和系统重采样技术防止粒子权重的衰退。仿真分析了高斯白噪声信道下的载波恢复性能，结果表明，基于粒子滤波的载波恢复算法载波恢复时间短，同步速度快，恢复频率范围广。建立以Watterson模型为基础的短波通信系统模型，通过该模型在不同的信道状态、信噪比和频偏条件下进行了仿真实验。结果表明在大频差的条件下，算法也具有良好的符号检测性能。

关键词: 短波突发通信, 无辅助数据, 粒子滤波