Plenary Lecture

Queuing and Loss Network Models:
Computational Algorithms and Asymptotic Analysis

Professor Hisashi Kobayashi
Princeton University, USA
E-mail: hisashi@Princeton.edu

Abstract: Queueing network theory has been successfully applied by computer and communication system modelers to represent the inherent contention and congestion in multiple resource systems, to identify the system bottlenecks, and to assess the performance limits. A queueing network model provides a suitable framework for analyzing the performance of “connection-less services” in a packet switched network. The so-called “product-form” networks such as Jackson network and its generalizations, allow such performance metrics as throughput and the mean delay to be represented by a ratio of the “normalization constants” with different arguments.

Connection-oriented services, such as the conventional circuit-switched telephone networks and end-to-end flow connections over the Internet can be properly represented by loss network models. The loss network theory is a relatively recent development, and can be viewed as an extension of the classical Erlang and Engset loss models.

We will discuss interesting relations between queueing networks and loss networks, and show that the computational algorithms developed for queueing networks are equally applicable to the normalization constants and performance metrics in loss networks as well.

Finally, we will discuss the case of large systems, where even most efficient algorithms for exact solutions are computationally infeasible. Recent development for approximation techniques and asymptotic performance limits will be reviewed.

References:
1. H. Kobayashi and B. L. Mark, System Modeling and Analysis, Pearson Prentice Hall, 2008
2. H. Kobayashi and B. L. Mark, “Product-Form Loss Networks,” in Frontiers in Queueing (ed. J. H. Dshalalow). pp. 147-195, CRC Press, 1997.
3. F. P. Kelly, “Loss Networks,” Ann. Appl. Prob., vol.. 1, no. 3, pp 319-378, 1991.

Brief Biography of the Speaker:
Hisashi Kobayashi is the Sherman Fairchild University Professor of Electrical Engineering and Computer Science at Princeton University since 1986, when he joined the Princeton Faculty as Dean of the School of Engineering and Applied Science (1986-91). Prior to joining Princeton he worked for the IBM Research Division for 19 years (1967-86). He was the founding director of IBM Tokyo Research Laboratory (1982-86). He received his BS (1961) and MS (1963) from Tokyo University and his MA (1966) and Ph.D.(1967) from Princeton. He was a radar engineer at Toshiba, Japan (1963-65).

His principal fields of research are system modeling and analysis, queuing theory and signal processing algorithms. He has also worked on data transmission theory, digital magnetic recording, optical network architectures, wireless geolocation algorithms, and network security. He was a recipient of the 2005 Eduardo Rhein Technology Award of Germany for his 1969 invention of a high-density digital recording scheme, now widely known as PRML (partial response coding, maximum likelihood decoding).

He is an IEEE Fellow (1977), IEEE Life Fellow (2003), and IEICE Fellow (2004). He received the Humboldt Prize of West Germany (1979) and IFIPS Silver Core (1980), and is a member of Japan’s National Academy of engineering (1992). He published “Modeling and Analysis” (Addison Wesley, 1978) , coauthored with Brian Mark a textbook “System Modeling and Analysis” (Pearson-Prentice Hall, 2008) and is currently working on “Probability, Random Processes and Statistical Analysis,” to be published by Cambridge University Press in 2009.