CNN Based Resource Management for D2D Networks with Wireless Networks Virtualization

The energy consumption of the system with wireless network virtualization will increase as the number of slices increase. Meanwhile, device-to-device (D2D) communication can reduce the load of base stations and improve spectrum efficiency, which applied in wireless virtualized networks can improve the system’s energy efficiency. However, D2D communication can cause interference to cellular users, how to ensure the communication quality of different users while keeping the system energy consumption lower is noteworthy. In this paper, the problem of wireless resource virtualization allocation with D2D communication is formulated. We formalized the channel allocation and power control problem with the goal of system energy efficiency. We solve this problem at a lower computational complexity with convolutional neural network (CNN). Results show that faster speedups are obtained with lower losses compared to optimal results by the proposed scheme.