Reinforcement Learning Approch Handover Management for Hybridlifi and Wifi Networks

Under ever increasing of demand for mobile data results to the looming crisis of a spectrum crunch in the existing wireless fidelity (WiFi) networks. To alleviate the spectrum crunch one possible solution is to exploit and investigate the convergence of different networks domain. In an indoor environment, integrating high speed-data transmission of light fidelity (LiFi) and large coverage of WiFi, namely a hybrid LiFi and WiFi networks (HLWNets) have drawn significant attention. Meanwhile, handover issues in HLWNets become challenging because of the complete overlap between LiFi and WiFi coverage areas. Additionally, mobile user devices (UDs) in optical LiFiattocells cause frequent handover while WiFi is vulnerable to traffic overload. In this paper, a downlink hybrid system using four LiFi APs and a WiFi APs is considered to enhance mobility of UDs and reduce handover overhead using reinforcement learning (RL) approach. The decision to perform handover is dictated and analyzed through different key performance parameters. Those performance matrices are the rate of change signal interference plus noise ratio (SINR), average throughput of UDs, handover rate, and speed of UDs. The Results show that the proposed RL approach provides a noticeable performance for various mobile speed scenarios compared to the existing handover algorithm .