Efficient polling cycle adaptive passive optical network for low-latency 5G fronthaul

Under the large and ubiquitous bandwidth requirements of the upcoming 5G era, new common public radio interface (CPRI) solutions (such as eCPRI) characterized by scale flexible, packet-based bandwidth reduction are receiving extensive attention. In this context, applying a passive optical network (PON) into the mobile fronthaul (MFH) has good prospects due to the features of low cost and wide coverage. However, the millisecond-level latency of the dynamic bandwidth allocation (DBA) procedure exceeds the harsh latency constraints of 5G MFH, and the uneven distribution of traffic caused by user mobility brings waste or congestion of MFH bandwidth resources. To meet the latency constraint and simultaneously improve bandwidth utilization, we propose a flexible time- and wavelength-division multiplexed passive optical network (TWDM-PON)-based MFH architecture, in which optical network units (ONUs) can be dynamically grouped to support efficient use of the bandwidth and adaptive polling cycle. Based on this architecture, the influence of the propagation time on latency is analyzed, and a joint bandwidth and latency constraint (JBL) grouping algorithm is proposed to obtain the appropriate ONU group for different network slice scenarios. The simulation results indicate that, in the case satisfying the latency constraint, the bandwidth utilization reaches 94% and above (maximum 99.5%). Simultaneously, within different slicing scenarios, the polling cycle can be increased by 98% on average (maximum 122%) or decreased by a maximum of 36μs. Compared with the non-grouping situation, the adaptive polling cycle lowers the performance requirement of the hardware device (such as a burst mode transmitter), expands the MFH coverage, and improves the channel utilization.