Monday, 8 October 2007 - 09:00 - 10:00
Reliable systems from unreliable components: Von Neumann-Shannon Revisited
By Iraj Saniee, Bell Laboratories, Alcatel-Lucent, USA
Abstract: The keynote involves 4 parts:
A) A short review of various connection-oriented protection/restoration schemes --
1+1, 1+N, shared mesh, p-cycles, etc.
B) A short review of packet-oriented protection/restoration schemes --
resilient packet ring, resilient packet mesh, OSPF/BGP re-computations and convergence.
C) Going back to the roots: Replication and redundancy theory -- Von Neumann and Shannon theories
for computer and telephone network reliability.
D) Recent extensions of C (above) at the application level involving splitting, replication and coding --
these are modern version of Von Neumann-Shannon theory and are revolutionizing modern communication
while having resiliency and reliability built into their fabrics, as in peer-to-peer networking and large-scale
content distribution networking. I'll give examples from distributed storage, fountain and network coding and their
Biography: Iraj Saniee is a Director in the Mathematical and Algorithmic Sciences Research Center at Bell Laboratories, Alcatel-Lucent, in Murray Hill, New Jersey, USA. He received his B.A. (Hon.) and M.A. (Hon.) degrees in mathematics and his Ph.D. in operations research and control theory, all from the University of Cambridge. At Bell Labs, the emphasis of research in Dr. Saniee’s department is on network design and mathematical modeling, analysis, and optimization of emerging processes in optical, wireless and data networks. His recent research interests have been on new modes of slotted communication in WDM networks, flow control mechanisms for QoS in multi-service networks, and randomized resource allocation and distributed optimization. Prior to Bell Labs he was a director in the Applied Research Area of Bell Communications Research, Morristown, NJ, USA. Recent technology transfers from his department include SRLG-disjoint routing in the distributed Optical Network Navigator System (LambdaUnite) and schedulers for data users in cellular wireless systems (CDMA DOr0). He is a past chair of the Telecom Section at INFORMS, a member of IEEE, a member of the IFIP Working Group 7.3 on Computer Performance Modeling and Analysis and serves on the editorial board of Operations Research. He led the 1994 runner-up team in the INFORMS’ Edelman International Prize in Operations Research for the design, development and implementation of a system known as SONET Toolkit and has published numerous articles in IEEE, and INFORMS journals and proceedings.
Tuesday, 9 October 2007 - 09:00 - 10:00
By Charles Colbourn, Arizona State University, USA
Abstract: Simple network models often underpin the analysis and synthesis of reliable networks, yet even simple models do not lead to efficient analytic techniques. As a result, combinatorial techniques have been extensively explored in order to approximate the network structure, or to approximate the reliability measure of interest, with the goal of providing useful analytic tools that can be efficiently calculated. Here we explore the uses – and also the abuses and misuses – of combinatorial models in practical reliability assessment. The objective is to understand what combinatorial models can and cannot tell us, to ensure that despite their inherent limitations, we use them when appropriate.
Biography: Charles Colbourn completed his PhD at the University of Toronto in Canada in 1980; he chaired the Department of Combinatorics and Optimization at the University of Waterloo (Canada), and held an endowed professorship at the University of Vermont, prior to joining Arizona State University. He is the author of the graduate text “The Combinatorics of Network Reliability”, and has written extensively on techniques for calculating, bounding, and approximating network reliability. His research interests focus on combinatorial structure and algorithms that exploit it. He serves on six editorial boards of international journals in these areas. In addition to research on network reliability in wired networks, his recent research explores protocols in optical, mobile ad hoc, and sensor networks as well as network testing.
Wednesday, 10 October 2007 - 09:00 - 10:00
by Dimitri Papadimitriou, Alcatel-Lucent, Belgium
Abstract: MPLS was designed as a Layer 3 forwarding paradigm. After a period where MPLS in-band OAM functionality has been nearly totally ignored, MPLS OAM and related techniques are currently taking the opposite direction. Henceforth, we assist to a progressive overload of the MPLS OAM functionality and related mechanisms. In turn, networks are becoming more complex to configure, operate and maintain; thus, MPLS OAM is inducing the inverse result from the initial objective (simplify network operations).
This paper will trace back the steps that lead from no OAM consideration until the current situation for IP/MPLS networks and provide an analytical overview of the MPLS OAM techniques. This paper will analyze where and when OAM mechanisms can be valuable in IP/MPLS network resiliency operations. For this purpose, quantitative estimations on OAM performance vs resiliency, robustness and scalability objectives will be presented.
From this analysis, techniques for combining adequately OAM functionality will be illustrated on representative examples.
Biography: Dimitri Papadimitriou joined the Alcatel CTO in 2000 as Expert Research Engineer the Network Architecture team of the Central Research Center (CRC) where he was in charge of the packet/optical networking architecture. In 2004, he joined the Packet-Transport Infrastructure (PTI) team of the Research and Innovation (R&I) of the Alcatel Corporate CTO department. Since 2006, he works for the Network Strategy Group (NSG) of the Alcatel Corporate CTO, as Network Strategy Director where he is in charge of the carrier grade packet technonologies and network architecture. His current areas of investigation are focused on IP and Ethernet networks including routing, traffic engineering and QoS mechanisms. He authored various technical papers on routing and recovery techniques for packet / multi-layer networks. He is also actively involved in the standardization activities of the IETF Routing Area. Dimitri Papadimitriou is a Distinguished Member of the Alcatel Technical Academy (ALTA).