Poster Abstract: Optimizing Handover Parameters by Q-learning for Heterogeneous RF-VLC Networks

Existing literature studying the access point (AP)-user association problem of heterogeneous RF-VLC networks either investigate quasi-static network selection, which results in outdated decision for highly mobile scenario, or only consider the dwell time for the vertical handover from VLC to RF, which means the handover from RF to VLC happens immediately once the condition is met. In this poster, we propose a flexible and holistic framework, that is running a self-optimizing algorithm at the centralized coordinator which resides in the LTE eNB and controls the handover parameters of all the VLC APs under the coverage of LTE eNB. Based on Q-learning approach, the algorithm optimizes the time-to-trigger (TTT) values for vertical handover between LTE and VLC according to historical signal-to-noise ratio (SNR) measurements. Simulation results based on real measurements reveal the significant throughput improvement by optimizing TTT under mobile scenarios and the capability of self-optimizing handover parameters.