Abstract: As a prerequisite of data detection, the frame timing synchronization has been become the hot topic and the key issue of the communication. However, the existing methods without adopting Compressive Sensing（CS） method have to meet the Nyquist rate, resulting in excessive power consumption and design difficulty for analog-to-digital converter. Furthermore, the correct synchronization probability needs to be improved when the compressive sensing method is adopted. To improve the correct synchronization probability, this paper introduces the model-based CS theory into frame timing synchronization, and a new improved method named model-based Compressive Sampling Matching Pursuit（CoSaMP） algorithm is proposed. In the proposed method, the synchronization metrics from the compressive sampling are reconstructed by utilizing model-based CoSaMP, and the start of frame is then estimated with the reconstructed metrics. Compared to the existing CS-based, analysis and simulation results show that the correct synchronization probability is improved.

Key words: frame timing synchronization, model-based compressed sensing, synchronization metrics, Compressive Sampling Matching Pursuit（CoSaMP）

摘要： 作为数据检测的必要前提，帧定时同步一直是通信领域的研究热点与难点。然而，现有非压缩感知帧同步方法必须满足奈奎斯特速率，导致过度功耗以及模拟—数字转换器（Analog-to-Digital Converter，ADC）设计难度；而采用压缩感知（Compressed Sensing，CS）帧同步方法的正确同步性能有待进一步提高。为此，将模式化压缩感知（Model-based Compressed Sensing）技术引入到帧定时同步中，提出了一种基于模式化压缩的采样匹配追踪（Compressive Sampling Matching Pursuit，CoSaMP）方法，用以重构压缩采样下的同步度量并完成帧同步。分析与仿真结果表明，相对于现已有的基于CS的帧定时同步方法，提出方法改善了帧定时同步的正确同步概率。

关键词: 帧定时同步, 模式化压缩感知, 同步度量, 基于模式化压缩的采样匹配追踪（CoSaMP）