基于动态流感知的接纳控制策略研究

Dynamic Flow Awareness Based Admission Control Scheme

流感知接纳控制使用调度机制测量的优先队列队长和公平速率作为链路状态信息实现隐式接纳判决。在对动态流感知进行分析的基础上,定义了优先队列的拥塞指数。在不同拥塞指数条件下,提出分别使用与流峰值速率相关的接纳概率对多业务进行区分接入判决,通过拒绝高速弹性流的接入达到对低速实时流的性能保护。推导了实时流阻塞率和弹性流平均吞吐量的理论公式。仿真结果显示相对于基于测量的接纳控制算法,提出策略具有较低的阻塞率和更好的平均吞吐量性能。

Flow-aware admission control performs implicit admission decisions using the link state parameters, i.e. priority queue length and fair rate measured by scheduling scheme. First of all, congestion index of priority queue is explicitly defined based on the analysis of dynamic flow-awareness. For the varying congestion index, admissions for services are differentiated by corresponding decisions with admission probability determined by the peak rate of flows. Thus the performance of real-time streaming flows can be guaranteed by reducing the admission of elastic flows. Furthermore, we deduce the formulas revealing the relationship between blocking rate of streaming flows and average throughput of elastic flows. Finally, the simnlating results show that the proposed scheme has lower blocking rate and better average throughput than measurement based on admission control algorithm.