摘要
针对多时分波分复用无源光网络资源缺乏全局性的集中控制和灵活分配的问题,建立了软件定义光接入网络架构,并提出一种带有预测的全局资源分配机制.基于软件定义光接入网络,运用模糊神经网络预测模型减小控制器与光线路终端信息交互时延,并采用自适应弹性权重对光线路终端间的资源进行分配;同时,对各光线路终端内的资源设计周期空闲时隙填充式的分配算法,进一步降低网络时延.仿真结果表明,所提资源分配机制从网络全局视角优化了不同层级资源的分配,有效地减小了网络整体时延,提高了信道利用率.
For lack of global centralized control and flexible allocation of the inter-TWDM-PONs network resource, the architecture of the software-defined optical access network is designed and a novel global resource allocation mechanism with prediction is proposed. Based on the software-defined optical access network, the fuzzy neural network predicting model is utilized to reduce the interactive information delay between controller and OLT. Simultaneously, the adaptive and flexible weighting is utilized to allocate the inter-OLTs resource. Furthermore, an allocation algorithm with filling idle slots is designed to further reduce the network delay for the intra-OLT resource. Simulation results show that the proposed mechanism can optimize allocation of the different layers resource from the global network view. Also, the overall network delay is effectively reduced, and the channel utilization is imt3roved.
作者
熊余
张振振
师劲
吴大鹏
XIONG Yu ZHANG Zhenzhen SHI Jin WU Dapeng(Key Lab. of Optical Communication and Networks, Chongqing Univ. of Posts and Telecommunications, Chongqing, 40006)
出处
《西安电子科技大学学报》
EI
CAS
CSCD
北大核心
2017年第5期140-146,共7页
Journal of Xidian University
基金
国家自然科学基金资助项目(61401052)
重庆市教委科学技术研究资助项目(KJ1400418
KJ1500445)
重庆邮电大学博士启动基金资助项目(A2015-09)
关键词
时分波分复用无源光网络
软件定义网络
模糊神经网络
全局资源分配
空闲时隙
time and wavelength division multiplexed passive optical network
software defined network
fuzzy neural network
giobal resource allocation
idle time
