Robust routing under dynamic traffic demands

Agrawal, H 2009, Robust routing under dynamic traffic demands, Doctor of Philosophy (PhD), Electrical and Computer Engineering, RMIT University.

Document type: Thesis
Collection: Theses

Attached Files
Name Description MIMEType Size
Agrawal.pdf Thesis application/pdf 2.40MB
Title Robust routing under dynamic traffic demands
Author(s) Agrawal, H
Year 2009
In order to provide service reliability with reasonable quality, it is essential for the network operator to manage the traffic flows in the core network. Managing traffic in the network is performed as routing function. In the traditional traffic management, network operator can tune routing parameters to simply manage the traffic. But traditional routing methods are not designed to handle the sudden fluctuations in the traffic. As a result, this may apparently lead to the traffic congestions in some parts of the core network, leaving other part underutilized. In this thesis we explore issues related to the routing robustness in the face of traffic demand variations.

We investigate different routing methods for efficient routing using maximum link utilization (MLU) as a performance metric. The primary advantage of using link utilization is its ease to compute the network performance on real network data and synthetic data. Overloaded links might result in Quality of Service degradation (e.g. larger packet delay, packet losses etc.), so MLU might be a useful measure of network performance. For the experimentation, we have used unique data from the real operational network available in the public domain and the random data for large network topology instances.

Furthermore, we propose a simple routing algorithm called Robust Routing Technique (RRT) to implement a robust routing mechanism. This mechanism allows network operator to satisfy the networking goals such as load balancing, routing robustness to the range of traffic demand matrices, link failures or to the traffic changes caused by uncertain traffic demands. Simulation experiments with real network topologies and random topologies demonstrate that our routing solution is simple (for routing) and flexible (for forwarding). K-Shortest path implementation in RRT can be extended for Multi Protocol Label Switching.

Finally, we evaluate the performance of robust routing under dynamic traffic demands. We formulate the problem as a multi commodity flow problem using linear programming. We use congestion ratio to define the robust routing performance. We provide a variant to the existing robust routing mechanisms by modelling traffic demand due to Distributed Denial of service attacks or worms. Simulation results are compared with the popular OSPF traffic engineering algorithm to provide effectiveness to the proposed routing scheme. Simulation results are compared with the popular OSPF traffic engineering algorithm to provide effectiveness to the proposed routing scheme.
Degree Doctor of Philosophy (PhD)
Institution RMIT University
School, Department or Centre Electrical and Computer Engineering
Keyword(s) Routing
traffic demand
Version Filter Type
Access Statistics: 246 Abstract Views, 486 File Downloads  -  Detailed Statistics
Created: Tue, 10 May 2011, 16:52:03 EST by Guy Aron
© 2014 RMIT Research Repository • Powered by Fez SoftwareContact us