Gradient Boosted Trees Based Mode Selection Decision for Moving D2D-Enabled Heterogeneous Ultra-Dense Networks

Abstract

With the evolution of fifth-generation (5G) communication systems toward to heterogeneous ultra-dense networks (H-UDNs). Device-to-device (D2D) communications have been proposed as a promising technology to improve system capacity and user experiences. However, in moving D2D-enabled H- UDNs, it will cause heavy system overhead from the frequent mode selection between D2D mode and cellular mode. In this paper, in order to achieve a trade off between the advantages of D2D communications and system overhead, we propose a Gradient Boosted Trees (GBT) based multi-attribute D2D mode selection decision strategy. The proposed strategy combines Received Signal Strength (RSS), the Signal to Interference plus Noise Ratio (SINR) and moving angle of users to assist base stations (BSs) in selecting the optimal communication mode for user equipments (UEs) in mode selection decision process. Simulation results show that our proposed strategy brings improvements to the mode selection performance, which can be reflected in reducing the mode selection probability and increasing the D2D mode dwell time. Moreover, the system overhead is further reduced and system throughput increases significantly.