基于改进模糊分类模型的光纤网络数据调度研究

Research on optical network data scheduling based on improved fuzzy classification model

针对光纤网络中由于色散导致的双向传输时延不对称,易产生数据锐变等问题,传统调度方法采用模糊聚类方法进行光纤数据信息属性聚类和调度,对数据聚集中心敏感、光纤数据个体间的子空间配对容易使数据访问过程陷入局部最优解,造成数据聚集延迟长、通信性能差的问题。提出一种基于改进模糊分类模型的光纤网络数据调度方法。在综合考虑性能指标、服务能力以及服务质量等指标的基础上,给出光纤数据的综合评价模型,将任务和网络资源构成混合矩阵进行数据格式化,并构造相应的光纤数据模糊近似度矩阵进行聚类,调度策略引入光纤数据紧急度、稀缺度来选取网络数据调度的优先级,并利用离散粒子群算法对整体的调度方案进行优化,找出最优数据调度策略集。实验结果表明,所提方法能够高效地实现光纤网络数据调度,且与其它方法比较,降低了数据聚集延迟,实现了整个光纤网络通信性能最优化。

Aiming at the asymmetry of bidirectional transmission delay caused by dispersion in optical fiber network,it is easy to produce data sharpening. The traditional scheduling method uses fuzzy clustering method to cluster and schedule fiber data information attributes. The pairing of subspaces of sensitive and between optical data individuals in data clustering center easily cause the data access process fall into a local optimal solution,resulting in a problem of long data aggregation delay and poor communication performance. A fiber optic network data scheduling method based on improved fuzzy classification model is proposed. Based on the comprehensive consideration of performance indicators,service capabilities and service quality,a comprehensive evaluation model of fiber data is given. The task and network resources are combined into a matrix to format the data,and the corresponding fiber data fuzzy approximation matrix is constructed to cluster and schedule strategy which introduce fiber data urgency and scarcity to select the priority of network data scheduling. Use discrete particle swarm optimization to optimize the overall scheduling scheme and find the optimal data scheduling strategy set. The experimental results show that the proposed method can efficiently implement fiber network data scheduling,and compared with other methods,the data aggregation delay is reduced,and the communication performance of the entire fiber network is optimized.