To date, the mainstream ethernet passive optical network (EPON) bandwidth allocation schemes as well as the new IEEE 802.3ah Ethernet in the first mile Task Force specifications have been centralized, relying on a component in the central office [optical line termination (OLT)] to provision upstream traffic. Hence, the OLT is the only device that can arbitrate time-division access to the shared channel. Since the OLT has global knowledge of the state of the entire network, this is a centralized control plane in which the OLT has centralized intelligence. This paper proposes novel distributed EPON architectures, and in the process proves that these distributed networking architectures and the associated bandwidth allocation algorithms and protocols have characteristics that make them far better suited for provisioning quality of service (QoS) schemes necessary for properly handling data, voice, video, and other real-time streaming advanced multimedia services over a single line. Specifically, this paper proposes a novel ethernet over star coupler-based PON architecture that uses a fully distributed time-division multiple-access (TDMA) arbitration scheme. Supported by the decentralized scheme, this paper proposes several QoS-based dynamic bandwidth allocation (DBA) algorithms in which the OLT is excluded from the implementation of the time slot assignment. In contrast with the mainstream centralized EPON architectures that combine priority queuing [intra-optical network unit (ONU) scheduling] with DBA schemes (inter-ONU scheduling), the proposed distributed EPON architecture supports differentiated services through the integration of both scheduling mechanisms at the ONU (intra-ONU scheduling). The introduction of this integration feature that can only be supported by a decentralized architecture provides better QoS guarantees. Furthermore, in addition to the added flexibility and reliability, the overall performance of the proposed decentralized EPON architecture and the associated bandwidth allocation algorithms are shown to be at least as efficient as their centralized counterparts.
Read full abstract