Research On EPON Grant Scheduling Algorithms And Auto-discovery Process Scheduling

Ethernet Passive Optical Network (EPON) is a promising and widely deployed broadband access technology. In an EPON system, how to allocate the limited upstream bandwidth among different Optical Network Units (ONU) is an important research topic. Static bandwidth allocation scheme allocates a fixed amount of bandwidth for each ONU. Although simple and with guaranteed bandwidth, the static bandwidth allocation scheme fails to dynamically adjust to the different bandwidth requests of different ONUs, and is comparatively low bandwidth efficient. In order to support statistic multiplexing, Dyanmic Bandwidth Allocaiton (DBA) scheme is proposed. And there are tremendous researches on this sheme considering different requirements. However, the study of grant scheduling in DBA which aims at minimizing the mean packet delay is relatively un-covered in the literature.At the same time, in order for the Optical Line Terminal (OLT) to discovery the newly switched-on ONUs and the reconnecting ONUs after failure, an auto-discovery process is defined in EPON standard. Because the auto-discovey process would comsume a part of the upstream bandwidth, it is meaningful to work out an optimal scheduling scheme for auto-discovery process to improve the bandwidth efficiency and lower the ONU registration delay.This paper will first briefly introduce the EPON system, and then focus on the DBA algorithms and auto-discovery process. Based on the system model information, this paper will intensively consider the grant scheduling algorithms and the scheduling optimization for auto-discovery process.In grant scheduling algorithms study, we analyze the relationship between the mean packet delay and grant scheduling algorithms. In order to support QoS guarantee and fairness, EPON system generally requires an offline scheduling framework. In such framework, The OLT will first collect bandwidth requests from all ONUs, and then make bandwidth allocation and scheduling decisions for the shared upstream channel. Due to varing Round-Trip Time (RTT) and grant window sizes, the transmission order of ONUs will greatly affect the mean packet delay. In order to optimize the transmission order for all ONUs and minimize the mean packet delay, this paper will first conduct a rigorious mathematic analysis on the impact of RTT and grant window size on mean packet delay. Under different RTT and grant window size condition, this paper proposes different scheduling strategies. Based on those strategies, this paper proposes an Integreated Grant Scheduling (IGS) algorithm. Simulation results show that the performance of IGS is better compared with other algorithms under various conditions.In the study for scheduling optimization of auto-discovery process, we first propose the scheduling optimization problem for auto-discovery process, and propose a novel simulation based scheduling optimization solution. This solution can provide guarantee for the registration delay of ONU and also can decrease the bandwidth usage for auto-discovey process. Previous studies mainly focus on the collision resolution mechanism. And candidate mechanisms are random delay, binary exponential back-off, and the combination of both. Previous studies have confirmed that random delay mechanism is the most bandwidth efficient and can achieve the lowest registration delay for un-registered ONUs. Howerer, there is few studies in the literature that study the scheduling algorithms under the random-delay mechanism. This paper will study the scheduling optimization under the random-delay mechanism. And the main concern is to determine the optimal frequency of auto-discovey process and the discovery window size.In conclusion, this paper introduces the principal technologies related to EPON standard, and mainly studies the grant scheduling algorithms and auto-discovery process scheduling in depth. The outcomes are a meaningful complement to the research area.