Dynamic Resource Allocation for Indoor VLC/WiFi Heterogeneous Network Based on Q-learning

The rapid increase in user data traffic has led to conventional single radio frequency (RF) wireless communication no longer meeting user demand. Light-emitting diode (LED) visible light communication (VLC) can also not meet the communication demand due to its over-reliance on line-of-sight transmission. The heterogeneity of visible light networks and traditional RF networks can meet user demand well. However, because the communication link limits LED optical communication, the traditional resource allocation algorithm can no longer meet user demand and system utilisation. This paper addresses the issue by proposing a joint method for channel detection and access based on Q-learning multi-agent and a dynamic strategy for bandwidth adjustment for fixed channels. This approach resolves the issue of multiple users frequently switching links. It optimises the VLC communication constrained by the communication link using the benchmark the user transmits in the optimal channel link. Simulation results show that the proposed strategy dramatically improves the system’s throughput, utilisation and user fairness compared with the traditional link-switching method with fixed bandwidth.