A learning algorithm for TWDM-PON' DWBA with 5G fronthaul networks

摘要

Time and wavelength-division multiplexed passive optical network (TWDM-PON) is expected to be the fronthaul networks of 5G, which requires low latency and large bandwidth. In this paper, we propose a Q-learning based dynamic wavelength and bandwidth allocation (DWBA) algorithm for TWDM-PON. Simulation results show that the proposed DWBA algorithm can reduce the number of active channels by 40% and provide larger bandwidth with same wavelength number in comparation with the existing algorithms while satisfying the latency requirement of 5G fronthaul networks.