ACM Home


Erol Gelenbe 
Department of Electrical and Electronic Engineering
Intelligent Systems and Networks Group
Imperial College, London

Tuesday, June 3, 8:45-9:15






Title: New Product Form Solutions for Stochastic Models: Discovery or Invention?

In the late seventies, most of us agreed that most of the queuing networks having product form solutions had already been discovered. Thus much research was then devoted to explaining why certain queuing networks had such solutions, and theories about “one step transitions”, the “M=>M property”, “local balance”, and “quasi-reversibility” were put forth. The G-networks that we introduced in the late 1980’s and early 1990’s, initially met with incredulity (in particular at the SIGMETRICS conferences), but proved in a series of papers [3,4,6,7] that one could have product form solutions (a) without local balance, (b) without one step transitions, (c) without quasi-reversibility, and (d) with non-linear traffic equations. These new models established the possibility of incorporating control primitives into solvable models of queuing networks. G-networks were inspired by neuronal signalling, and resulted in interesting parallels between neuronal models [1,2,8], gene regulatory networks [9] and queuing networks. We were able to show that (a) they approximate continuous and bounded functions, and (b) that they have gradient optimisation algorithms which are O(n3) for an n-queue network [5,8]. In this lecture we will briefly review these results and summarise recent product form results that we have discovered.

[1] E.  Gelenbe ``Random neural networks with positive and negative signals and product form solution'', Neural Computation, 1 (4): 502-510, 1989.
[2] E.  Gelenbe ``Stability of the random neural networks'', Neural Computation, 2 (2): 239-247, 1990.
[3] E.  Gelenbe ``Product form queueing networks with negative and positive customers'', J. App. Prob., 28: 656-663, 1991.
[4] E.  Gelenbe ``G-networks with instantaneous customer movement'', J. App. Prob. , 30 (3): 742-748, 1993.
[5] E.  Gelenbe ``Learning in the recurrent random network'', Neural Computation, 5: 154-164, 1993.
[6] Erol Gelenbe ``G-Networks with signals and batch removal'', Probability in the Engineering and Informational Sciences, 7: 335-342, 1993.
[7] E. Gelenbe, J.M. Fourneau ``G-Networks with resets'', Performance Evaluation, 49: 179-192, 2002.
[8] E. Gelenbe, K. Hussain ``Learning in the multiple class random neural network'', IEEE Trans. on Neural Networks 13 (6): 1257-1267, 2002.
[9] E. Gelenbe ``Steady-state solution of probabilistic gene regulatory networks'', Physical Review E, 76(1), 031903 (2007).

Erol Gelenbe (FACM, FIEEE, FIEE) has been characterized as the single individual who, over a span of 30 years, has made the greatest overall contribution to the field of Computer System and Network Performance Evaluation through research, doctoral training, wide ranging international collaboration, and professional service. He holds the “Dennis Gabor Chair” at Imperial College, and his papers in 2007-2008 appear in the ACM Trans. on Sensor Networks, ACM Transactions on Internet Technology, Neural Computation, Performance Evaluation, Physical Review, Proceedings of the Royal Society A, and The Computer Journal.  He has made decisive contributions to product form networks by inventing G-networks (Gelenbe-Networks) with totally new types of “negative customers, triggers, and resets”, and which are characterised by non-linear traffic equations. He has made seminal contributions to random access communications, the optimisation of reliability in database systems, the design of adaptive QoS-aware packet networks, diffusion models in performance analysis, and the performance of link control protocols.

A native of Turkey who graduated from the Middle East Technical University in Ankara, he has authored four books written in English and French, two of which have appeared in Japanese and Korean translation, and over 115 journal articles in the Journal of the ACM, Physical Review, Acta Informatica, Proceedings of the Royal Society, Management Science, IEEE Trans. on Computers, IEEE Trans. on Neural Networks, IEEE Trans. on Software Engineering, IEEE Trans. on Systems Man and Cybernetics, IEEE J. on Selected Areas in Communications,  ACM Trans. on Sensor Networks, Communications of the ACM, Journal of Applied Probability, and Theoretical Computer Science, etc..  His recent work includes path finding algorithms in noisy and uncertain conditions, networked auctions, the use of neural networks to control routing in computer networks, as well as theoretical biology and theoretical chemistry including the analysis of neural networks and gene regulatory networks. His research is currently funded by industry (GD, BAE Systems and QinetiQ), and by agencies including EPSRC, MoD and DoD, and the EU.

A founder of IFIP WG7.3 and of ACM SIGMETRICS, and of the journal Performance Evaluation, Erol is particularly proud of the 58 PhD students he has graduated, many of whom are prominent in academia and industry in France, the USA, Turkey, Greece, UK, Canada, Belgium and Venezuela. Appointed to a chair at the age of 27 at the University of Liège in Belgium, he founded performance modelling at INRIA and in French Universities. His professorial posts include the University of Paris-Orsay, the University of Paris-Descartes, New Jersey Institute of Technology, Duke University and the University of Central Florida. He is now Editor-in-Chief of The Computer Journal (British Computer Society), and serves on the editorial board of several journals. Erol is currently a member of the Science and Technology Board and of the Executive Board of the UK Defence Technology Centre on Data and Information Fusion. His experience includes being Department Head at Duke University (USA), Associate Dean at the University of Central Florida (USA), and chairing the Technical Advisory Board of the US Army’s Simulation and Training Command (1999-2003). 

His honours include: Commander of Merit of the Republic of Italy, Grand Officer of the Star of Italy, Officer Order of Merit of France and Chevalier des Palmes Académiques. Member of the French National Academy of Engineering, Turkish Academy of Sciences, Academia Europaea. He received the Science Award (1994) of the Parlar Foundation in Turkey, was the first computer scientist to be awarded the Grand Prix France Telecom (1996) of the French Academy of Sciences, and received “honoris causa” doctorates from the University of Rome II (1996), Bogaziçi University, Istanbul (2004), and the University of Liège, Belgium.

Harrick M. Vin
Vice President of Research and Development at Tata Consultancy Services

Tuesday, June 3, 13:00-14:00










Title: Navigating Complexity through Managed Evolution

The complexity of large-scale “information technology (IT) plants”—consisting of a number of hardware and software components—has been increasing rapidly and is fast approaching a barrier. Continuous evolution is a key contributor to this complexity. IT plants evolve to accommodate new software functionality, hardware technology, application and user requirements, as well as changes in operating conditions (workload, faults, etc.). To navigate the complexity of IT plants, we need methodology and
tools for managing evolution. These tools should facilitate designers and operators of IT plants to reason about the impact of each evolution step on system-level properties—such as security, privacy, performance, availability, and reliability--and thereby simplify many tasks such as capacity management, performance engineering, system testing, and real-time decision making (e.g., anomaly detection, alert generation, and root cause analysis). In this talk, I will briefly describe the overall R&D vision, some of the key challenges, our approach, and some of the early results from analyzing performance and operational data collected from large-scale operational systems.

Dr. Harrick Vin is a Vice President (R&D) and a member of the Corporate Technology Board at Tata Consultancy Services (TCS), India. TCS is the largest Asian IT Services company. At TCS, he heads the Systems Research Lab (SRL) -- responsible for developing innovative solutions for next-generation enterprise infrastructure. Prior to TCS, he was a Professor of Computer Sciences at the University of Texas at Austin. His research interests are in the areas of networks, operating systems, distributed systems, and multimedia systems. Harrick received his Ph.D. in Computer Science from the University of California at San Diego. He has co-authored more than 100 papers in leading international journals and conferences. He is a recipient of several awards including the Faculty Fellow in Computer Sciences, Dean's Fellowship, National Science Foundation CAREER award, IBM Faculty Development Award, Fellow of the IBM Austin Center for Advanced Studies, AT&T Foundation Award, National Science Foundation Research
Initiation Award, IBM Doctoral Fellowship, NCR Innovation Award, and San Diego Supercomputer Center Creative Computing Award. He has served on the Editorial Board of ACM/Springer Multimedia Systems Journal, IEEE Transactions on Multimedia, and IEEE Multimedia. He has been a guest editor for IEEE Network. He has served as the conference and program chairperson for the premier ACM and IEEE international conferences in the area of multimedia systems and networks; and served as a technical program committee member for many international conferences.

François Baccelli
Professor, INRIA (Unité de Recherche de Rocquencourt) and ENS (Département d'Informatique) France

Wednesday, June 4,  8:30-9:30










Title: Measurement Based Self-Optimization of Wireless Networks using Gibbs Fields


The popularity of IEEE 802.11 WLANs has led to dense deployments, either managed or unmanaged, in urban areas, enterprises and campuses.  High density leads the close proximity of co-channel cells, increased interference and contention and hence to sub-optimal performance. An efficient use and sharing of the spectrum requires some form of network-level organization.

This talk surveys a collection of fully distributed self-organization algorithms applicable within this context
and which allow:

  1. multiple interfering 802.11 Access Points to select their operating frequency in order to minimize interference throughout the network;

  2. users to choose the Access Point they attach to in order to optimally share the global network bandwidth;

  3. Access Points to control their transmission power in order to optimally share the global network bandwidth. This power control problem is particularly challenging as  variable transmit powers result in asymmetric links in the network and can potentially lead to throughput starvation of some nodes.

The proposed algorithms all rely on the Gibbs sampler and its annealed version, which can be seen as distributed algorithms for minimizing some global energy function. We analytically prove their convergence: Algorithm (i) leads to minimal total interference and Algorithms (ii) and (iii) lead to minimal total potential
delay (when considering elastic traffic on the downlink).

These algorithms do not require global coordination among the wireless devices. They only require the participating wireless nodes to measure such local quantities as interference or transmission delay and to perform some local adaptive tuning of the relevant parameters:  Access Point frequency for (i), user Access Point attachment for (ii) and Access Point transmit power and carrier sensing parameter for (iii).

We study their performance using both simulation and testbeds and we show that their incremental deployment is feasible. This class of algorithms is versatile and can be adapted to various other contexts. To illustrate this, some recent extensions to optimal routing in mesh WLANs will also be discussed.

This survey is based on a series of papers presented at Infocom 07 (on user association, frequency selection and on power control) and at CoNext 07 (on routing) which are  based on collaboration  with Vivek Mhatre (Motorola), Bruno Kauffmann (INRIA-ENS), Augustin Chaintreau (Thomson),  Christophe Diot (Thomson), Konstantina Papagiannaki (INTEL) and Henrik Lundgren (Thomson).

Francois Baccelli is a specialist of communication network modeling and design.

He got his "doctorat d'etat" from Universite Paris-Sud in 1983. He held positions at INRIA Rocquencourt, AT&T Bell Laboratories and INRIA Sophia Antipolis. He is currently INRIA "directeur de recherche" in the computer science department of Ecole Normale Superieure in Paris, where he started the research group on communication networks in 1999.

He is the co-author of a book on queueing theory with P. Bremaud and on a book on the max plus algebra with G. Cohen, G.J. Olsder and J.P. Quadrat. His current research work is focused on two topics: 1)  the analysis, the control and measurements of large IP networks and 2) the development of new stochastic geometry tools for assessing and exploiting the capacity of wireless networks.

F. Baccelli is a member of the French academy of sciences.

Jeannette Wing
Assistant Director, Computer and Information Science and Engineering Directorate, National Science Foundation

Thursday, June 5, 14:00-15:00










Title: Computational Thinking and Thinking about Computing

My vision for the 21st Century: Computational thinking will be a fundamental skill used by everyone in the world. To reading, writing, and arithmetic, let's add computational thinking to every child's analytical ability. Computational thinking has already influenced other disciplines, from the sciences to the arts.  Realizing this vision gives the field of computing both exciting research opportunities and novel educational challenges.

The field of computing is driven by technology innovation, societal demands, and scientific questions.  We are often too easily swept up with the rapid progress in technology and the surprising uses by society of our technology, that we forget about the science that underlies our field.  In thinking about computing, I have started a list of "Deep Questions in Computing," with the hope of encouraging the community to think about the scientific drivers of our field.  One of these questions, “How can we build complex systems simply?” is of special relevance to the SIGMETRICS community.  I will close my talk by posing some research challenges specific to measuring properties of complex systems.

Dr. Jeannette M. Wing is the President's Professor of Computer Sciencein the Computer Science Department at Carnegie Mellon University. Shereceived her S.B. and S.M. degrees in Electrical Engineering and Computer Science in 1979 and her Ph.D. degree in Computer Science in 1983, all from the Massachusetts Institute of Technology. From 2004-2007, she was Head of the Computer Science Department at Carnegie Mellon. Currently on leave from CMU, she is the Assistant Director of the Computer and Information Science and Engineering Directorate at the National Science Foundation.

Professor Wing's general research interests are in the areas of specification and verification, concurrent and distributed systems, and programming languages. Her current focus is on the foundations of trustworthy computing.

Professor Wing was or is on the editorial board of eleven journals. She has been a member of many advisory boards, including: the Networking and Information Technology (NITRD) Technical Advisory Group to the President's Council of Advisors on Science and Tecbnology (PCAST), the National Academies of Sciences's Computer Science and Telecommunications Board, ACM Council, the DARPA Information Science and Technology (ISAT) Board, NSF's CISE Advisory Committee, Microsoft's Trustworthy Computing Academic Advisory Board, and the Intel Research Pittsburgh's Advisory Board. She is a member of the Sloan Research Fellowships Program Committee. She is a member of AAAS, ACM, IEEE, Sigma Xi, Phi Beta Kappa, Tau Beta Pi, and Eta Kappa Nu. Professor Wing is an AAAS Fellow, ACM Fellow, and IEEE Fellow.

Last updated:06/30/2008