决策域流感知接纳控制

针对单信息流感知接纳控制存在的判决误差，提出使用优先队列队长和链路公平速率指标构建4个二维状态空间以描述不同的链路状态信息。根据不同链路状态下的流分布特征，将4个状态空间映射为3个联合决策域，并由此得到多业务流的联合判决表作为提出接纳控制算法的判决条件。针对不同业务流各自的敏感性能指标，推导了实时流的阻塞率和弹性流的平均吞吐量表达式，仿真统计显示提出算法在过载情况下，实时流阻塞率平均增加约5.44%时，其时延减少约为总时延的10%，弹性流吞吐量平均增加为链路带宽的25.75％。

Decision Domain Based Flow-Aware Admission Control

To avoid the decision error of single information based flow-aware admission control, four 2D state spaces are established using priority queue length and link fairness rate to describe the information of different link states. According to the flow distribution under different link states, four state spaces are mapped into three integrated decision domains. The integrated decision table based on the integrated decision domain is the decision conditions for proposed admission control algorithm. The formula of blocking probability for real-time flows is derived, and the average throughput for elastic flows obtained. Simulation results show that the proposed algorithm can reduce the delay of real-time flows by 10% and increase the average throughput of elastic flows by 25.75% with approximately 5.44% increase of the blocking probability for real-time flows when overloaded.