论文部分内容阅读
针对资源约束网络负载的动态变化,设计了一个基于最小二乘支持向量机(LSSVM)的反馈调度器.它可以周期性地监测网络资源,在线预测下一周期的可适用网络利用率,并根据预测值采用插值法得到控制回路的下一个采样周期,从而实现系统资源的动态分配.对采用固定带宽分配、基于LSSVM以及基于Elman神经网络的反馈调度进行了比较,结果表明,所提出的策略能使系统在可变负载情况下稳定运行,并在控制质量和网络服务质量之间取得平衡.
Aiming at the dynamic change of resource constrained network load, a feedback scheduler based on least square support vector machine (LSSVM) is designed, which can periodically monitor the network resources and predict the available network utilization rate in the next cycle online. The predicted value is interpolated to obtain the next sampling period of the control loop to realize the dynamic allocation of system resources.Compared with fixed bandwidth allocation, feedback scheduling based on LSSVM and Elman neural network, the results show that the proposed strategy can Make the system run stably under the condition of variable load, and strike a balance between control quality and network service quality.