Millimeter wave (mm-Wave) communication is a practicable scheme for big data communication in next generation cellular communication. Since mm-Wave frequencies have an extremely large path loss, the large antenna arrays are still in small signal-to-noise ratio before beamforming, which leads that acquiring channel knowledge to design these beamformers is challenging. In this paper, we concentrates on using support information extracted at Sub-6GHz to aid the beam selection in a multi-user mm-Wave system, which adopts a Hybrid analog/digital beamforming structure. We formulate mm-Wave beam-selection as a compressive sensing problem, and outline a corresponding strategies, such as constructed a structured beamformer design and a weighted sparse signal reconstruction, which are depend on Sub-6GHz spatial information. The simulation results for achievable rate show that the proposed out-of-band aided beam-selection strategies can reduce the training overhead compared with conventional strategies.