Abstract: Many systems are characterized by multiscale features in nature, and the systems are frequently observed and measured in different scales. Complex networks can be regarded as the refined model for describing complex systems. The multiscale geometric analysis provides a powerful microscopy to observe the structures of complex systems. Consequently, complex networks can be analyses via multiscale geometric analysis. Keeping in mind this ides, a compression strategy based on multiscale geometric analysis is proposed for depicting the structure of complex networks with the small amount of information. The analysis and the experiments indicate that the proposed method is feasible, and performs as a sparsely representing for successfully understanding and compressing networks, which can be useful for analyzing networks with the small amount of information.