一种基于C-RAN架构下资源负载预测方法

在C-RAN构架下,虚拟基带池集中化资源管理要求对物理硬件资源进行实时监控,保证物理资源有效进行动态分配和负载均衡。提出将自回归AR(autoregressive)预测方法运用于C-RAN架构下的物理资源负载预测,并根据物理资源特点,构建预测模型。采用双阈值限定负载范围,对超过或低于阈值的后续时间节点的负载进行预测,避免瞬时负载峰值触发的虚拟基站迁移。仿真结果说明该方法可以减少迁移次数,达到整个系统中资源的负载均衡。

A Prediction Method of Resource Load based on C - RAN Architecture

Abstract] For C - RAN architecture, the centralized resource management in virtual baseband pool requirements real-time monitoring of the physical hardware resources, so that to ensure effective dynamic distribution and load balancing of physical resources. In this paper, the autoregressive AR (Autoregressive) prediction method is used in load prediction of physical resources under the C-RAN architecture, and according to the features of physical resources, this paper constructs a forecast model. Using double thresholds to limit load range. And to predict the load of the subsequently-time nodes when it above or below the threshold. It avoids instantaneous load peak triggering migration of the virtual base stations. Experimental result shows that this method could reduce the number of migration to achieve the purpose of load balancing in the system.