Paris-Networking

About Paris-Networking | Announce a talk | Subscribe

Thesis Defence: On Selfish and Malicious Behaviors in Wireless Networks: A Non-cooperative Game Theoretic Approach  

Lin CHEN, TELECOM ParisTech

Monday, November 3rd 2008, 10h00 - 13h00

Location :

Room B312,
TELECOM ParisTech,
46, rue Barrault,
75013 Paris.
M° Corvisart.

Abstract :

Examination Board :
- Pr. Tamer Basar (reviewer, University of Illinois at Urbana-Champaign),
- Pr. Ken Chen (reviewer, University of Paris 13),
- Pr. Jean-Pierre  Hubaux (reviewer, EPFL),
- Pr. Kaldoun Al Agha (University of Paris-Sud),
- Pr. François Bacelli (INRIA/ENS),
- Pr. Philippe Godlewski (TELECOM ParisTech),
- Dr. Jean Leneutre (supervisor, TELECOM ParisTech).

Abstract:
In this thesis, we present an axe of research where non-cooperative game theory
is applied as a framework to model and analyze selfish and malicious behaviors
in wireless networks. More specifically, the following selfish and malicious
behaviors are systematically studied:
* Selfish behaviors
- MAC layer selfish behaviors in IEEE 802.11 wireless networks
- Non-cooperative power and rate control in IEEE 802.11 wireless networks
- Cooperative transmission in non-cooperative wireless networks
* Malicious behaviors and defense strategies
- Intrusion detection in heterogenous networks
- Jamming attacks and defense strategy
- Multihop routing amid malicious attackers

By employing non-cooperative game theory as a line of research, for each
selfish/maliciousbehavior, we formulate the corresponding non-cooperative game
in the specific context of that problem. The resulting Nash equilibrium (NE) is
then derived, followed by an analysis on the key properties of the game
solution, i.e., the existence, uniqueness of the NE, the convergence to the NE
and the efficiency of the system at the NE.  Concerning selfish behaviors, this
analysis serves as foundations for the further design of incentive-compatible
protocols and pricing mechanisms to fill the gap between the inefficient NE and
the global optimal or quasi-optimal point. In the study of malicious behaviors,
this analysis leads to the development and validation of new defense mechanisms
seeking to eliminate the unfavorable NE from the defender's perspective if
multiple NEs exist and limit the damage caused by malicious attackers at the
remaining NE.

Host :

Jean Leneutre - TELECOM ParisTech