On June 12, 2015, Niek Baër will defend his PhD thesis “Queueing and Traffic”. Prior to this ceremony, a seminar is held. We cordially invite you to attend both the symposium and the PhD-defence. The event will be held at the University of Twente.
Date&time: June 12, 9:45 – 17.30
Location of Mini Symposium: Waaier 3
Location of the Defense: Berkhoff Auditorium
9.45 - 10.15 Welcome coffee and tea
10.15 - 10.30 Opening by Richard Boucherie
10.30 - 11.15 Eric van Berkum (University of Twente)
11.15 - 12.00 Bo Friis Nielsen (Technical University of Denmark)
12.00 - 13.00 Lunch (Waaier)
13.00 - 13.30 Peter Kovacs (University of Amsterdam)
13.30 - 14.00 Ahbishek (University of Amsterdam)
14.30 - 14.45 Introduction to the defence
14.45 - 15.30 PhD defence of Niek Baër
15.45 - 16.15 Ceremony
16.30 - 17.45 Reception
10.30 – 11.15 Eric van Berkum (University of Twente):
11.15 – 12.00 Bo Friis Nielsen (Technical University of Denmark):
Distributions with rational moment generating functions
13.00 – 13.30 Peter Kovács (University of Amsterdam):
Proportionally fair scheduling for traffic light networks
Abstract: We consider a decentralized traffic signal control policy for urban road networks. The policy uses a proportionally fair scheme to allocate green times at each junction using only local information of queue lengths. At the same time it maintains a cyclic ordering of service phases which is a desirable property for traffic safety. Our model also includes elements of real life traffic behaviour such as the use of switching times and the time dependency of service rates due to the initial setup at the beginning of service phases. We discuss the effect of these phenomena on the set of feasible traffic demands. We also investigate the impact of the chosen cycle lengths on the ensuing waiting times of the vehicles. As our main result, we show that the proportionally fair policy stabilizes the network for any feasible demand. In order to accomplish this, we construct the fluid limit of the underlying stochastic system and give a formal proof of convergence as a generalization of prior results in the context of telecommunication applications. Existing stability results in the literature for proportional fair scheduling cannot be directly applied to our model because of several characteristics that are specific to the application to road traffic networks, including the earlier mentioned cyclic service patterns and the time dependency of service rates.
13.30 – 14.00 Abhishek (University of Amsterdam):
Queueing analysis of a highway ramp with varying traffic density
Abstract: We investigate a flow of cars onto a highway with varying traffic density. Our aim is to characterize the impact of the variability of the traffic density on the capacity and the delay of cars on the ramp. Our model is a queueing system with multiple service types and correlated service times having general distribution functions. The variability in the service times reflects the varying traffic density.
Specifically, we assume that there are N different service types (i.e., density profiles on the highway), each having their own (arbitrary) service time distribution. The "service type" of a car depends on the service type and the service duration of the previous car.
We investigate this queueing system in equilibrium and find the joint distribution of an arbitrary car's service duration and service type by determining the partial generating functions. The solution technique entails the derivation of a linear system of equations for the partial generating functions and then determining a set of unknown coefficients by employing Rouché's to determine the number of zeros of a determinant inside the unit disc.