“…For instance, the authors in [4] propose an algorithm that divides the total upstream bandwidth into fixed-sized bandwidth units (say 10 Mbps per bandwidth unit), such that, high-priority ONUs receive more bandwidth units than best-effort ones. Further refinements are proposed in [5], [6], [7] where the ONUs have several Virtual Output Queues (VOQs) for different traffic classes, such that the granted transmission window is shared by the ONU's traffic classes following some weighted algorithm that assigns more time to high-priority VOQs than to low priority ones. Essentially, in most cases, it is the OLT which guarantees a fair sharing of bandwidth between the ONUs, while each ONU partitions the granted transmission window to the traffic classes following some weighted algorithm [8].…”