Description:

Project kind: theoretical analysis/simulations

For some communication scenarios (e.g. multiple access) we now know at which rates it is possible to communicate reliably. These rates depend on system parameters such as the power and the bandwidth of individual users. Individual users may have a clear preference regarding the desired outcome (e.g. their desired rate) and may set the parameters that they control (e.g. power) in such a way as to benefit to the detriment of other users. One can use concepts and tools from microeconomics and game theory to better understand how users make decisions, how they trade, as well as to understand how one can provide incentives that lead to outcomes that are socially desirable.

Objective:
We take a look at the multiple access AWGN channel. The users in this channel are considered rational: their only goal is to maximize their own satisfaction. We model the satisfaction of user i with any given operating point (in terms of power P and rate R) with a utility function u_i(P, R). We know that, under very strong conditions, the only way to efficiently use the communication resources is to make the users pay a tax, which we call the “truth tax”. This tax changes the incentives for the users such that their selfish behavior actually leads to the best outcome for the society. The obtained result is valid if we assume that each user tries to maximize his utility in each “turn”. However, if we relax this assumption, and assume that users want to maximize their average utility over multiple turns (and accept to possibly be suboptimal in any single turn), then efficient utilization of resources can be achieved more easily. The objective of this project is to investigate the behavior of users under these relaxing assumptions, and come up with the set of requirements for the efficient utilization of resources.

A prospective student is expected to have basic knowledge of Information Theory and Probability Theory. The work on this project will consist of learning some game theory (interest and previous knowledge of this field is helpful, but not required), then using it to model a communication scenario, working out some analytic results, and, time permitting, doing some simulations in Matlab. As a consequence of working on this project, the student can expect to learn basics of Game Theory and Optimization, as well as to consolidate their knowledge of different fields mentioned here by applying them to a specific communication setting.

References:
[1] Andreu Mas-Colell, Michael D. Whinston, and Jerry R. Green. Microconomic Theory. Oxford University Press, 1995.

Prerequisites:
Information Theory
Probability Theory
Game Theory (optional)

Supervisor:
Vojislav Gajic, LCM, Tel.: 33147, Room: INR 031, vojislav.gajic@epfl.ch

Professor:
Bixio Rimoldi

back to master semester projects menu