基于自相似排队模型的星上交换输入缓存分配算法

Input-buffer allocation algorithm for satellite switching system based on self-similar queuing model

针对空间信息网络的自相似分组业务流、星上交换节点的有限缓存资源,提出一种基于自相似排队模型的星上交换输入缓存分配算法。构建了基于VOQ的空间信息网络星上Crossbar交换节点模型,依据空间信息网络业务流的自相似性,通过建立虚通道自相似排队模型估算出各个虚通道的缓存溢出概率,再根据其溢出概率采用贪婪算法来实现缓存资源的优化分配。仿真分析了在自相似分组业务流下优化分配算法和均匀分配算法的性能。结果表明,在相同缓存开销下,优化分配算法可以有效降低分组的平均时延;在同等分组平均时延下,分组业务流自相似程度为0.8和0.6时,可分别节省缓存开销约25%和17.4%。

For the self-similar packet traffic in space information network(SIN)and the finite buffer resources in satellite switching node,this paper proposed an input-buffer allocation algorithm for satellite switching system based on self-similar queuing model.Firstly,it established the model of the satellite Crossbar switching node in SIN based on VOQ.According to the self-similar traffic properties of SIN,it estimated the buffer overflow probability of each virtual channel through the queuing model of the virtual channels under the self-similar traffic.And then it used greedy algorithm to optimize allocation of buffer resources.Finally,it analyzed the characteristics of optimal allocation algorithm and uniform allocation algorithm by simulation.The results show that,for the same buffer consumption,the optimal allocation algorithm is more effective to decrease average packet latency.On the other hand,for the same average packet latency,the optimal allocation algorithm can save 25%and 17.4%of buffer consumption when the self-similarity of packet traffic is 0.8 and 0.6.