Soutenance de Thèse de Ahmed Farhan HANIF
« Resource Utilization Techniques in Distributed Networks with Limited Information »
Mercredi 7 mai 2014 à 9H45 en salle A003
à Télécom SudParis (ex Télécom INT)
(9 Rue Charles Fourier, 91000 Évry)
As systems are becoming larger, it is becoming difficult to optimize them in acentralized manner due to insufficient backhaul connectivity and dynamical systems behavior. In this thesis, we tackle the above problem by developing a distributed strategiclearning framework for seeking Nash equilibria under state-dependent payofffunctions. We develop a discrete time stochastic learning using sinus perturbation with the realistic assumption, that eachnode only has a numerical realization of the payoff at each time. We examinethe convergence of our discrete time algorithm to a limiting trajectory defined byan ordinary differential equation (ODE). Finally, we conduct a stability analysisand apply the proposed scheme in a generic wireless networks. We also providethe application of these algorithms to real world resource utilization problems inwireless. Our proposed algorithm is applied to the followingdistributed optimization problems in wireless domain. Power control, beamforming and Bayesian density tracking in the interferencechannel.
We also consider resource sharing problems in large scale networks (e.g. cloudnetworks) with a generalized fair payoff function. We formulate the problem as a strategic decision-making problem (i.e. a game). We examine the resourcesharing game with finite and infinite number of players. Exploiting the aggregatestructure of the payoff functions, we show that, theNash equilibrium is not an evolutionarily stable strategy in the finite regime.Then, we introduce a myopic mean-field response where each player implementsa mean-field-taking strategy. We show that such a mean-field-taking strategy isevolutionarily stable in both finite and infinite regime. We provideclosed form expression of the optimal pricing that gives an efficient resource sharingpolicy. As the number of active players grows without bound, we show thatthe equilibrium strategy converges to a mean-field equilibrium and the optimalprices for resources converge to the optimal price of the mean-field game. Then, we addressthe demand satisfaction problem for which a necessary and sufficiency conditionfor satisfactory solutions is provided.
Composition du jury :
- M. Luis MUNOZ, Rapporteurs, University of Cantabria, Spain
- M. Rahim TAFZOLLI, Rapporteurs, University of Surrey, UK
- M. Pierre SENS, Examinateur (Président), Université Paris 6, France
- M. Pascal BIANCHI, Examinateurs, Telecom ParisTech, France
- Ms. Nidhi HEGDE, Examinateurs, Technicolor Labs, France
- M. Mohamad ASSAAD, Co-encadrants, Supelec, France
- M. Hamidou TEMBINE, Co-encadrants, KAUST, Saudi Arabia
- M. Djamal ZEGHLACHE, Directeur de thèse, Telecom SudParis, France