Myopic model of resource sharing in a queuing network

Revues récentes

Myopic model of resource sharing in a queuing network
Revue des ISTs de Madagasikara
auteur : T.B. RAVALIMINOARIMALALASON ; M. RAKOTOMALALA
Index terms : Game theory, Network, Queue, Resource, Sharing

[ENG] Resources in a queuing network are often rare and expensive. Sharing them to customers of the network is a hard problemand impacts a lot the performances of the system. We present in this paper a new optimal and feasible method to sharethese resources based on the result of the game theory. We use of the theory of bargaining game and repeated stochasticgame to find an optimal consensus for resources sharing based on customer aspirations. Our model, which we call “myopicmodel”, proposes a system of myopic players who don’t project future payoffs of the game, and which is a rather dynamicmodel depending on the evolution of customer aspirations over the time. Simulations of our hypothesis were performed byanalyzing performances of our system compared to First-In First-Out queue, standard Processor Sharing model, and oneusing Kalai-Smorodinsky bargaining solution. The results have allowed us to appreciate ours, mostly in an instabilityphenomenon, and to propose a variety of applications, mainly in any system with shared resources, such as distributedsystems, centralized systems as cloud computing, Search Engine Optimization algorithms, ...
Télécharger