I. Introduction
With the popularization of 5G communications, more and more new mobile applications have increasingly high requirements on mobile networks in terms of speed and latency. Meanwhile, the rapid development of communication technologies also requires flexible heterogeneous network convergence [1]. Millimeter-wave can provide a considerable bandwidth (more than 2 Gbps peak speeds). However, because of the limited radiation range inherent in mmWave, the current use of mmWave networks is very local. Therefore, the self-organizing heterogeneous mmWave networks can offer more flexible, complementary, full-coverage network service capability. Furthermore, it also helps mmWave networks eliminate heavy dependence on infrastructure, especially for mobile edge computing scenarios such as Internet of things (IoT) and vehicle-to-everything (V2X) communications. On the other hand, the smaller radiation coverage of mmWave also means more potential spatial spectrum holes in high-density mobile networks. Exploring and making full use of these potential spectrum holes through cognitive radio technologies will im-prove radio spectrum efficiency.