Research Initiation: Scheduling Policies for Real-Time Communications

下一个十年将看到越来越多的一体化水平， 在一个共同的网络中满足各种通信需求。 一直 一直对分组交换网络感兴趣， 实现这一目标。 虽然数据流量的性能 以延迟、实时流量（如语音和 视频以分组丢失来测量。 除了给 实时流量优先于数据流量，几乎没有工作 在确定调度策略时， 丢包 该研究将网络中的节点建模为 带有客户（数据包）的队列，每个客户都有一个截止日期 如果超过，则导致分组丢失。 一 调度策略，称为最短灭绝时间（STE） 策略，被证明是最佳的一类队列。 这些 结果被扩展到更大类的离散系统， 包括多服务器和多类队列。 以帮助 分析和设计具有这种调度策略的队列， 开发了计算这种队列的损失的方法。

The next decade will see increasing levels of integration of all kinds of communication needs in a common network. There has been a marked interest in packet-switched networks as a means to attain that goal. While the performance of data traffic is measured in terms of delay, real-time traffic such as voice and video is measured in terms of packet losses. Besides giving real-time traffic priority over data traffic, little work has been done in determining scheduling policies which minimize packet losses. This research models nodes in the network as queues with customers (packets), each of which has a deadline for service which, if exceeded, leads to a packet loss. A scheduling policy, called the shortest time to extinction (STE) policy, is shown to be optimal for a class of queues. These results are extended to a larger class of queueing systems, including multi-server and multi-class queues. To aid the analysis and design of queues with such scheduling policies, methods of computing losses for such queues are developed.