CAJ | 학술논문

基于Erlang-A同Erlang-C相结合的排队系统模型，由多云域组成的分散型云计算被抽象为多排队系统。基于丢弃率参数实现了对延迟敏感服务请求的准入控制，等待的服务请求以一定的丢弃概率被溢出。将丢弃概率引入Riordan公式，在Hayward近似中提出基于空间随机过程的汇聚溢出流渐进峰度因子，利用等效随机理论研究了溢出流丢失问题。通过仿真实验验证了基于丢弃率准入控制下多排队系统的性能提升。
기우Erlang-A동Erlang-C상결합적배대계통모형，유다운역조성적분산형운계산피추상위다배대계통。기우주기솔삼수실현료대연지민감복무청구적준입공제，등대적복무청구이일정적주기개솔피일출。장주기개솔인입Riordan공식，재Hayward근사중제출기우공간수궤과정적회취일출류점진봉도인자，이용등효수궤이론연구료일출류주실문제。통과방진실험험증료기우주기솔준입공제하다배대계통적성능제승。
Based on Erlang-A and Erlang-C ,a decentralized cloud computing system composed of multiple cloud domains is simplified to multiple queuing systems .The admission control of delay-sensitive service requests can be implemented based on aban-donment rate ,since waiting service requests are overflowed at a certain abandonment probability .The abandonment probability is in-troduced into Riordan Formula ,while the asymptotic peakedness factor of aggregate overflow based on spatial stochastic process is introduced into Hayward’s approximation .The overflow loss performance is studied using equivalent random theory .The perfor-mance improvement of the multiple queuing systems is testified by the numerical simulation under admission control based on aban-donment rate .