文中运用随机网络演算中统计流量包络和统计服务曲线表征网络中实时分组到达和服务的随机变化,提出概率可调度性的新概念,刻画某些弹性实时应用对QoS延时要求,通过研究某一时间间隔到达网络系统的累积分组的最坏情况延时,测试其是否满足相应的概率可调度性条件来实现实时QoS保证性能评价.概率可调度性是与实时分组到达时间相关的,因此,提出的方法能实现实时QoS保证下的瞬时特性研究以及研究某一时间间隔累积到达分组的概率可调度性.文中研究了FIFO网络系统中指数分布和一般分布服务时间下的分组概率可调度性特征,基于理论结论的数值结果和基于模型的模拟结果是一致的.同时,数值和模拟结果表明概率可调度性在实时分组的QoS保证性能评价中具有较好的适应性.
In this paper, the authors describe stochastically change of arrival and serving of realtime packets by a statistical traffic envelope and statistical service curve. To characterize the requirement of some elastically real-time applications for delay, a novel concept named probabilistic schedulability is proposed. The probabilistic schedulability is related to time t at packets arrival, so instantaneous characteristic can be analyzed and cumulative packets would be scheduled at any given real-time interval by the method. The paper investigates the probabilistic schedulability of real-time packets under exponential distribution and general distribution service time in a FIFO network. The numerical analysis based on the theoretic results is consistent with the simulation analysis. The results show that the method has more flexibility to characterize the schedulability of real-time packets.