Short bio. Lulai Zhu obtained a BEng and an MEng in Measurement Technology and Instruments from Sichuan University, China, in 2008 and 2011 respectively. After that, he joined Emerson Network Power and had been working as a software test engineer for four years. He received in 2016 an MSc in Computing Science from Imperial College London, UK, and is now a doctoral student in the Department of Computing under the supervision of Dr. Giuliano Casale. His PhD study aims to contribute analytical methods for solving performance models such as queueing networks and Petri nets, with a particular focus on fluid approximation. He recently worked as a research assistant in the DICE project on performance modeling and simulation, and would play a similar role in the RADON project on architectural decomposition and optimization. He is also an active maintainer of the well-known JMT suite.
Thesis abstract. As a multi-class variant of the classical egalitarian processor-sharing (EPS) discipline, discriminatory processor sharing (DPS) provides a suitable approach to model time-sharing operating systems and packet-switched communication networks where share exists to control the service. Although DPS is a more general scheduling discipline than the EPS, queueing models with DPS are of non-product form and thus far less mathematically tractable than those with EPS. A fluid approach is proposed in my PhD thesis for transient and steady-state analysis of closed QNs comprising delay and DPS queueing stations. By this approach, the expected sample path of a closed QN with DPS can be approximated using the solution to a system of ordinary differential equations, which is admitted to be unique in any finite time horizon.