A Novel Machine Learning-Based Handover Scheme for Hybrid LiFi and WiFi Networks

Abstract

Combining the high area spectrum efficiency of light fidelity (LiFi) and the ubiquitous coverage of wireless fidelity (WiFi), hybrid LiFi and WiFi networks have drawn increasing research attention. Meanwhile, the handover issue in hybrid networks becomes a hotspot since the coverage areas of LiFi and WiFi overlap each other. In addition, LiFi may cause frequent handovers for fast-moving users, while WiFi is susceptible to traffic overload. Consequently, the selection between LiFi and WiFi becomes a tricky problem. In this paper we propose a novel handover scheme, which adopts a dynamic coefficient via machine learning to adjust the selection preference between LiFi and WiFi. The new method balances channel quality, resource availability and user mobility to make handover decisions. Results show that compared to the received signal strength (RSS)-based and trajectory-based handover methods, the proposed scheme can improve the user's throughput by up to 260% and 50%, respectively.