US-Turkey Cooperative Research: System-Theoretic Approach to Teletraffic Analysis and Engineering

项目描述：该项目支持密苏里-堪萨斯城大学（UMKC）计算机科学与电信（CST）教授Khosrow Sohraby博士和土耳其安卡拉比尔肯特大学电气与电子工程系Nail Akar博士的合作研究。调查人员将研究应用于电信网络的流量建模和分析以及流量工程。排队系统是电信网络中的基本系统，而通信量是其中的主要组成部分。如果将排队系统的流量和服务过程建模为马尔可夫过程，那么在大多数情况下，排队系统的解可以归结为一组结构化马尔可夫链的解。这两位科学家引入了一种新的范式“广义不变子空间方法”，用于数值求解一组结构化的马尔可夫链，即所谓的M/G/1型马尔可夫链。他们计划将该方法的适用性扩展到更广泛的马尔可夫链集，用于从流体队列到随机服务器，从有限容量队列到连续时间队列的数值求解。他们还计划引入一种使用“最优控制”的跨学科技术，为综合当前互联网基础设施工程中使用的路径分裂机制的理论奠定基础。范围：该项目使两名具有互补专业知识的合格科学家能够解决在互联网和无线数据传输管理中重要的问题。这两位科学家在1994年至1996年期间进行了合作，并取得了重大成果。该项目将使他们能够结合这一多学科研究领域的专业知识，生产一个非常通用的引擎，解决通信网络性能评估领域出现的大多数马尔可夫系统。预计将有一名或多名比尔肯特大学的研究生参加UMKC的研究。

0115779 SohrabyDescription: This project supports collaborative research by Dr. Khosrow Sohraby, Professor of Computer Science and Telecommunications (CST), University of Missouri-Kansas City (UMKC), and Dr. Nail Akar, Department of Electrical and Electronics Engineering, Bilkent University, Ankara, Turkey. The investigators will study traffic modeling and analysis and traffic engineering as applied to telecommunication networks. Queuing systems are the fundamental systems in telecommunication networks, of which traffic is the major constituent. Solution of queuing systems in most cases reduce to the solution of a certain set of structured Markov chains if the traffic and server process are modeled as Markovian processes. The two scientists have introduced a new paradigm "Generalized Invariant Subspace Approach," which was used to numerically solve a set of structured Markov chains, the so-called M/G/1-type Markov chains. They plan to extend the applicability of the approach to a wider set of Markov chains, used in numerical solution of queues ranging from fluid queues to random servers, and from finite capacity queues to continuous-time queues. They also plan to introduce an interdisciplinary technique using "optimal control" to set the foundation for a theory to synthesize path-splitting mechanisms that are used in engineering of current Internet infrastructures.Scope: This project enables two qualified scientists with complementary expertise to address a problem that is important in the management of Internet and wireless data transmission. The two scientists collaborated in the 1994-1996 period and produced significant results. This project will allow them to combine expertise in this multidisciplinary research field to produce a very generic engine that solves most Markovian systems that arise in the field of performance evaluation of communication networks. One or more graduate students from Bilkent University are expected to join the research at UMKC.