长距离无源光网络中动态带宽分配算法研究

发布时间：2018-06-29 03:45

本文选题：长距离无源光网络 + 动态带宽分配　；参考：《大连海事大学》2016年硕士论文

【摘要】：长距离无源光网络(Long Reach Passive Optical Network,LR-PON)简化了网络结构,能够以较低成本为更多用户提供宽带接入服务,是下一代无源光网络技术的重要发展方向。LR-PON中光线路终端(Optical Line Terminal,OLT)与光网络单元(Optical Network Unit,ONU)之间的传输距离扩展到100kmm甚至更远,导致传播时延增加。为了充分利用上行信道带宽,减少上行数据包传输时延,如何设计有效的动态带宽分配算法(Dynamic Bandwidth Allocation,DBA)已成为LR-PON中迫切需要解决的关键问题之一。本文首先介绍了LR-PON网络的发展由来、网络结构和研究现状,随后对LR-PON中几种主要DBA算法进行了归纳和总结,分析了已有算法的优点和不足,然后针对LR-PON中ONU与OLT之间的传播时延差异性较大的特点,在多线程轮询调度算法(Multi-thread Polling,MTP)的基础上,考虑如何减少传播时延差导致的上行信道空闲时间,提出一种改进的多线程轮询调度算法(Improved Multi-thread Polling,IMTP).该算法依据ONU与OLT的传输距离将ONU分为长距离ONU和短距离ONU,针对从短距离ONU上传数据结束到OLT收到长距离ONU的第一个数据包之间的空闲时间,在短距离ONU之间依据其带宽请求比例进行分配,从而可以减少上行信道空闲时间,提高上行带宽利用率,降低上行数据包时延。其次,针对MTP算法采用固定的线程数目,无法依据网络负载的变化动态调整以提高信道利用率的局限性,提出一种基于上行数据包时延的自适应多线程轮询算法(Adaptive Multi-Thread Polling,AMTP)。在AMTP算法中,OLT根据上行数据包时延情况动态调整使用的线程数目,在低负载条件下如果上行数据包时延过大,则增加轮询周期的线程数目,否则减少线程数目：在高负载条件下,若上行数据包时延过大,则减少轮询周期的线程数目。这样可以进一步提高上行信道利用率,降低上行数据包时延。最后,通过OPNET仿真软件搭建LR-PON仿真平台,对IMTP和AMTP算法进行仿真研究,并与文献中已有算法进行对比。仿真结果表明：所提出的算法具有较小的平均上行数据包时延和数据包时延抖动以及较高的上行信道利用率,优于已有算法。
[Abstract]:Long reach passive Optical Network (LR-PON) simplifies the network structure and provides broadband access to more users at lower cost. LR-PON is an important development direction of the next generation passive optical network technology. The transmission distance between optical line terminal and optical network unit extends to 100kmm or more, resulting in the increase of propagation delay. In order to make full use of uplink channel bandwidth and reduce uplink packet transmission delay, how to design an effective dynamic bandwidth allocation algorithm (DBA) has become one of the key problems in LR-PON. This paper first introduces the origin, network structure and research status of LR-PON network, then summarizes several main DBA algorithms in LR-PON, and analyzes the advantages and disadvantages of existing algorithms. Then considering the difference of propagation delay between ONU and OLT in LR-PON, on the basis of Multi-thread polling scheduling algorithm (MTP), how to reduce the idle time of uplink channel caused by propagation delay difference is considered. An improved Multi-thread polling scheduling algorithm (IMTP) is proposed. According to the transmission distance between ONU and OLT, the algorithm divides ONU into long distance ONU and short distance ONUN, aiming at the idle time between the end of uploading data from short distance ONU and OLT receiving the first packet of long distance ONU. In order to reduce the idle time of uplink channel, improve the utilization of uplink bandwidth and reduce the delay of uplink data packet, the short distance ONU can be allocated according to the proportion of bandwidth request. Secondly, aiming at the limitation that the Adaptive algorithm adopts a fixed number of threads and can not dynamically adjust the network load to improve the channel utilization, an adaptive multi-thread polling algorithm (Adaptive Multi-Thread polling algorithm) based on uplink packet delay is proposed. In the AMTP algorithm, OLT dynamically adjusts the number of threads used according to the uplink packet delay, and increases the number of threads in the polling period if the uplink packet delay is too large under low load conditions. Otherwise, reduce the number of threads: under the condition of high load, if the uplink packet delay is too large, then reduce the number of threads in polling period. This can further improve the utilization of the uplink channel and reduce the uplink packet delay. Finally, the LR-PON simulation platform is built by OPNET, and the IMTP and AMTP algorithms are simulated and compared with the existing algorithms in the literature. Simulation results show that the proposed algorithm has smaller average uplink packet delay, packet delay jitter and higher uplink channel utilization than the existing algorithms.
【学位授予单位】：大连海事大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TN929.1

【相似文献】