Learning the Spectrum Using Collaborative Filtering in Directional Millimeter Wave Networks

Millimeter wave (mmWave) has triggered lots of new technologies for the next generation of wireless networks. However, learning the utility and the statistics of spectrum becomes difficult due to the directional antenna beams in mmWave networks. Due to the spatial correlation of spectrum, the efficiency of learning the spectrum can be improved by letting secondary users collaborate and exchange information. In this paper, the collaborative spectrum sensing in mmWave networks is studied by leveraging the similarity between the collaborative learning for spectrum and the recommendation system of electronic commerce such as Amazon. The technique of user-based collaborative filtering is applied. Real data is collected in the 60GHz band by capturing the data stream of the Dell WiGig DOCK and TP-Link WiGig Router. Numerical simulation shows that the proposed strategy can significantly improve the performance of spectrum sensing, especially for easing the false sensing caused by the rotations of antennas. It provides insight for the future study on the potential of spatial reuse in the mmWave band.