Description
Date depot: 15 décembre 2021
Titre: Making Cloud Energy-Efficient using Machine Learning and Game Theoretic Optimization
Directrice de thèse:
Maria TROCAN (LISITE)
Encadrante :
Shohreh AHVAR (Nokia-Bell-Labs)
Domaine scientifique: Sciences et technologies de l'information et de la communication
Thématique CNRS : Programmation et architecture logicielle
Resumé: The importance of cloud computing is growing rapidly in various platforms; IT, business and mobile applications resulting in increased number of users requests. Therefore, it requires continuous running of computational
devices to provide effective services and achieve user demands. However, long executions of computational devices can results in enormous amount of heat, that can cause hotspots resulting in extensive amount of energy
consumption. Its is important to control the heat of the cloud servers through proper thermal aware resource
scheduling in order to avoid hotspots. Furthermore, improper thermal management may decrease the reliability,
efficiency of the servers and cooling mechanism may cause high amount of cost. These issues can be addressed
through an intelligent thermal aware resource scheduling techniques and tuning cooling unit parameters. Therefore, in this PhD project considering the thermal heterogeneity of resources and cooling systems costs, the
resource allocation and scheduling problem will be modeled using game theoretic approach such as stackelberg,
Bayesian and non-cooperative game and predictive Machine Learning techniques to achieve balanced workload
among computing nodes.
Doctorant.e: Fahimullah Muhammad