Utility Max-Min Fair Resource Allocation for Diversified Applications in EPON

In EPONs, differentiated services enable higher quality of service (QoS) for some queues over others. However, owing to the coarse granularity of DiffServ, DiffServ in EPONs can hardly facilitate any particular QoS profile. This paper investigates an application-oriented bandwidth allocation scheme to ensure fairness among queues with diversified QoS requirements. We first define application utilities to quantify users' quality of experience (QoE) as a function of network layer QoS metrics. We then formulate the fair resource allocation issue into a max-min utility problem, which is quasi-concave over queues' delayed traffic and dropped traffic. We further employ the bisection method to obtain the optimal solution of the quasi-concave maximization problem. The optimal value can be achieved by proper bandwidth allocation and queue management schemes in EPONs.