==================
PhD position #1: Dynamic reconfiguration of IP/MPLS networks
===============================================
Nowadays, the management of an IP/MPLS network requires the use of
multiple specialized tools. The teams that are in charge of operating
the network use different tools for the management of equipment
configurations, traffic monitoring and network supervision. These tools
provide a certain ' view ' of the state of the network at a point in
time, but require the expertise of network specialists to decide
potential network adaptations.
The project NEC (/Network Engineering & Control/) aims at designing and
developing a complete environment for the operational management of
IP/MPLS networks. Beyond the simple supervision, what is targeted is the
' intelligent ' control of computer networks. It amounts to conceiving
and developing a Supervision/Control center that gives a global and
accurate view of the network state, with consistent and high-level data,
and that is a true engineering and decision-support tool based on
end-to-end stochastic simulation models and on optimization algorithms
allowing to take the optimal decisions. In addition to gathering all the
information at a centralized point, this tool will enable the proactive
analysis of all possible scenarios and the anticipation of the optimal
decisions in case of adverse events (QoS degradations, attacks,
failures, etc.).
The research of the PhD student will concern on one hand the estimation
of the network traffic matrix from SNMP link loads, and on the other
hand robust optimization algorithms for its dynamic reconfiguration. In
particular this research will involve the dynamic optimization of OSPF
weights and the rerouting of Label Switched Paths (LSP).
PhD position #2: Online simulation of IP/MPLS networks
==========================================
Nowadays, the management of an IP/MPLS network requires the use of
multiple specialized tools. The teams that are in charge of operating
the network use different tools for the management of equipment
configurations, traffic monitoring and network supervision. These tools
provide a certain ' view ' of the state of the network at a point in
time, but require the expertise of network specialists to decide
potential network adaptations.
The project NEC (/Network Engineering & Control/) aims at designing and
developing a complete environment for the operational management of
IP/MPLS networks. Beyond the simple supervision, what is targeted is the
' intelligent ' control of computer networks. It amounts to conceiving
and developing a Supervision/Control center that gives a global and
accurate view of the network state, with consistent and high-level data,
and that is a true engineering and decision-support tool based on
end-to-end stochastic simulation models and on optimization algorithms
allowing to take the optimal decisions. In addition to gathering all the
information at a centralized point, this tool will enable the proactive
analysis of all possible scenarios and the anticipation of the optimal
decisions in case of adverse events (QoS degradations, attacks,
failures, etc.).
The research of the PhD student will target analytic performance models
for an accurate online estimation of the quality of service of network
flows. One of the goals is to reduce the current gap between
packet-level models (that describes how TCP shares the bandwidth of a
network between persistent flows) and flow-level models (that describe
the dynamic at the session level), but also to integrate the streaming
traffics in the models. The proposed approaches will be based on
queueing theory, and in particular on GPS (/Generalized Processor
Sharing/) queueing networks.
PhD position #3: Large-scale simulation and dynamic reconfiguration of
cloud computing infrastructures
===========================================================================
We are witnessing a convergence between mobile phones and personal
computers. However there are still significant differences in the
underlying business models. The current model in mobile telephony
facilitates the access of inexperienced users to services/applications,
witha business model based on the billing for the purchase or use. In
contrast, for personal computers, the current model still assumes that
the user is able to manage its computer in terms of operating system,to
buy, install and updatesoftware modules, with the difficulties this may
represent for a newbie. In a professional environment, it is of course
possible to resort to specialists, but this requires substantial human
resources. This is conducive to the emergence of a new business model
for personal computers that is inspired by mobile telephony in terms of
service and use. This new business model should be based on the
incredible success of remote access models tosoftware resources (SAAS,
cloud computing), but also on the new possibilities offered by autonomic
computing and virtual machines.
The project SOP aims at designing a hybrid model for personal computers,
targeted both at indivual users and at professional users, and that is
able to combine the advantages of a local installation/configuration of
software tools with the advantages of the remote accesses to software
services hosted by a "data center" or by the personnal computer of
another user in a cycle-stealing mode.
The research of the PhD student will be done two parts. The first part
will aim at modelling the global system using stochastic models based on
queueing theory. The goal is to evaluate its performances when the
number of users is in tens or hundreds of thousands. The second part of
the thesis will be devoted to the design and mathematical analysis of
distributed algorithms for dynamic load-balancing and virtual machine
migration. This second part will be based on theoretical tools coming
from non-linear optimization theory and game theory.
Deadline: ASAP
Contact information: brun@laas.fr
**********************************************************
*
* Contributions to be spread via DMANET are submitted to
*
* DMANET@zpr.uni-koeln.de
*
* Replies to a message carried on DMANET should NOT be
* addressed to DMANET but to the original sender. The
* original sender, however, is invited to prepare an
* update of the replies received and to communicate it
* via DMANET.
*
* DISCRETE MATHEMATICS AND ALGORITHMS NETWORK (DMANET)
* http://www.zaik.uni-koeln.de/AFS/publications/dmanet/
*
**********************************************************
Posts
