中文摘要：

设计了一种由2个广义等分结构的正交频分复用（OFDM）符号组成的新训练序列,并基于该序列提出一种新颖的OFDM同步方法.该方法用2个训练符号生成的多个相关峰合并,形成符号同步序列.在较低的计算复杂度条件下,可获得较好的符号时间估计性能.用2次低复杂度的延时自相关产生的相关值的相角,经合并后,获得2个初始频偏估计值.利用数论方法,组合2个估计值,实现较大范围的频偏估计.由于采用的是时域估计方法,无需对整个训练序列作快速傅里叶变换,故具有计算复杂度较低的优点.仿真结果表明,当本方法选取合适的广义等分数L时,能够估计出较大范围的频偏,并获得较好的频偏估计性能.

英文摘要：

A new synchronization scheme was proposed based on a new training sequence which is composed of two OFDM symbols with wide sense repetition structure. After combining several auto-correlation peaks from two OFDM training symbols, this algorithm maintains a good estimation performance of time under a low system complexity. After combining the phase from auto correlation with different delays, two initial estimations of the frequency offset corresponding to the two training symbols can be obtained. According to some knowledge on number theory, those two initial estimations jointly estimate the frequency offset with a quite large range. This algorithm operates in time domain, thus, it involves no FFT computation of the whole training sequence, leading to a low system complexity. The computer simulations show that by appropriately choosing the repetition number L, the scheme is able to deal with a relatively large range of frequency offset, while maintaining a good estimation performance.