在低频射电波段观测宇宙黑暗时期和再电离时期是未来10-20年观测宇宙学最前沿的关键突破方向之一。如何在这一新观测窗口内准确识别和分离强度相差5个量级以上的各层次天体辐射成分，并最终找到来自上述两个时期高度红移的极微弱中性氢21cm信号，需统合物理、天文、数学、信号分析与统计学等多学科知识，是当今国际重大课题。面对这一挑战和机遇，我们建议以上海交大低频射电理工团队为基础，发挥其十年来积累的优势，开展以下工作，为21世纪第二个十年观测宇宙学的进步做贡献，为未来全面开展宇宙再电离和黑暗时期研究提供重要观测分析手段：（1）基于多波段信息在角分和亚角分尺度上实现高精度低频射电天空仿真模拟；（2）据此，提出强前景干扰条件下探测效果稳定可靠的微弱信号识别和提取新算法，其有效探测阈值在图像空间至少达到10-3量级、在频谱和功率谱空间至少达到10-5量级；（3）将新优化算法用于实测数据及相关天体物理研究。

The dark ages and its late stage, i.e., the epoch of reionization (EoR), are expected to be a major center of interest within the next one or two decades for the facilities working in the low-frequency radio band, such as 21CMA, LOFAR, MWA, and SKA. One of the serious challenges in observing the 21 cm signals from the dark ages and EoR is that they are extremely weak and have been redshifted to the low frequency radio band. We propose to simulate the low frequency radio sky by taking into account different celestial emission components to achieve reasonably higher spatial and spectral accuracies. Base on the sky simulation new signal separation algorithms are to be designed to guarantee that different emission components can be identified and separated correctly. Also we will apply these new signal separation algorithms to the observed images and spectra to constrain the study of galaxies, galaxy clusters, and the the Epoch of Reionization.