I am interested in various topics in queueing theory. My PhD project focussed on single-server queues with server unreliability.
I have also worked on polling and priority systems, numerical queueing analysis techniques, queues with M/G/∞ input, paired queueing systems, etc. In addition,
I have studied various queueing systems, inspired by applications in telecommunications, health care scheduling and inventory management.
I work on applications of branching processes in queueing theor and epidemiology. In queueing theory, the dynamics of various polling systems can be described by branching processes.
Branching processes can further be used to describe the initial phase of an epidemic (compartmental or on a network).
I am interested in stochastic epidemic models and their fluid limits for information propagation in social networks.
I am currently working on network epidemics with Koen De Turck, Balakrishna Prabhu and Konstantin Avrachenkov and on compartmental models with Koen De Turck and William Knottenbelt.
I work/have worked on analytic models for various components of wireless networks including energy saving protocols in WiMax and LTE networks,
energy harvesting sensor nodes, information propagation in ad-hoc networks, opportunistic scheduling, etc.
I am mainly interested in tractable stochastic models for assessing the performance of such networks.
Health care scheduling
I work on fast and accurate numerical evaluation techniques for outpatient schedules. Such techniques are useful for optimising the schedules according to performance criteria such as patient waiting times, physician idle times, session over-time, etc. In addition, fast evaluation methods can be used to provide online patient waiting time estimation.
As physicians usually see but a few patients in a session, steady state queueing analysis does not apply. Moreover, patients may be unpunctual or may not show up at all.
This is joint work with Stijn De Vuyst.
Optical buffering is based on fibre delay lines (FDLs) as optical random access memory (RAM) does not exist.
The optical packet is "stored" by sending it through a delay line.
The length of the FDL and the entry time of the packet completely determine when the packet leaves the FDL.
I.e., FDL lines hold the information for a fixed amount of time as opposed to RAM where one may retrieve the stored information at any time.
Moreover, an FDL set cannot generate just any delay which leads to capacity loss. I have worked on queueing analysis of optical buffers with Wouter Rogiest, Koenraad Laevens, Evsey Morozov, Benny Van Houdt and Joris Walraevens.
Links to the outlines of the courses are below. Course notes and slides are available through Minerva.
Queueing analysis and Simulation (with Joris Walraevens)
Simulation of Manufacturing and Service Systems (with Stijn De Vuyst)
Stochastic processes for Electrical Engineering
Computer Intensive Statistical Methods
People I work with
Eitan Altman at INRIA Sophia-Antipolis, France
Konstantin Avrachenkov at INRIA-Sophia-Antipolis, France
Majed Haddad at INRIA-Sophia-Antipolis, France
Balakrishna Prabhu at LAAS-CNRS, Toulouse, France
William Knottenbelt at Imperial College, London, UK
Evsey Morozov at Petrazavodsk University, Russia
Stijn De Vuyst at Ghent University
Koen De Turck at Ghent University
Wouter Rogiest at Ghent University
Jan-Pieter Dorsman at Eindhoven University of Technology
Eline De Cuypere (co-supervised by Koen De Turck)
Ekaterina Evdokimova (co-supervised by Sabine Wittevrongel)
Hannah Van den Bossche (co-supervised by Herwig Bruneel)
Research Foundation Flanders - Stochastic modelling and performance analysis of WiMAX networks (2011-2015)
Research Foundation Flanders - Modelling and control of energy-harvesting wireless sensor networks (1/1/2015-31/12/2018)