Toward parallel edge computing in long-reach PONs

The huge bandwidth capacities and low costs of passive optical networks (PONs) combined with their high data rates have made them strong candidates for wireless backhauls. Many designs have therefore been proposed to integrate PONs with edge and fog computing paradigms, which are essential for many emerging applications. However, the feasibility of this integration has not yet been fully examined. The dynamic bandwidth allocation (DBA) that would best support this integration and the effect it would have on network performance have not yet been studied. In this paper, we study the performance of edge computing in long-reach PONs (LR-PONs), where long propagation delays pose challenges to the bandwidth allocation performance. We believe this paper is one of the first to study the feasibility of edge computing in these optical access networks by investigating whether centralized or decentralized allocation would be better to support computational offloading to the edge. We compare centralized multithread polling and a modified decentralized scheme in terms of offloading delays, effects on upstream traffic delays, and throughput.