Projet de recherche doctoral numero :2917


Date depot: 1 janvier 1900
Titre: Schémas d'allocation de ressources distribués pour réseaux cognitifs
Directeur de thèse: Philippe GODLEWSKI (LTCI (EDMH))
Directeur de thèse: Marceau COUPECHOUX (LTCI (EDMH))
Domaine scientifique: Sciences et technologies de l'information et de la communication
Thématique CNRS : Non defini

Resumé: The proposed research consists in designing and evaluating the performances of distributed resource allocation algorithms for cognitive networks. The notion of cognitive radio comes from the observation that the radio spectrum is not efficiently utilized. There is indeed a high heterogeneity of its usage both in time and space. FCC measurements have shown that large parts of the spectrum remains unused although allocated, while in the same time, some spectrum bands are subject to congestion because of the increasing demand for wireless services. An important challenge for current research in the wireless networks community is to propose efficient algorithms to cope with this issue. This is facilitated by the rapid development of software radio, which allows radio equipments to be agile and flexible. They can indeed rapidly change frequency band, sense or scan several channels or even recon gure protocols. In this context, Mittola has developed the notion of cognitive radio, which is particularly adapted to the primary/secondary usage of the spectrum. In this model, a primary user (PU) is the license owner of a spectrum band, possibly divided into several frequency channels, and has a strict priority for using this resource for its own transmissions. It can be a TV broadcaster or a cellular network operator. This primary user is however not using radio resource at 100%, either because transmissions are sporadic in the time domain or because it leaves some geographical white spaces unused. This offers the opportunity for secondary users (SU), which are also cognitive radios (CR), to opportunistically access the spectrum and so make a better usage of it. A central constraint for SUs is to not disturb in anyway primary users transmissions. The way secondary users choose the best frequency channels, access to the medium and share the resource with other cognitive radios is related to radio resource allocation schemes. These schemes are by nature distributed thus should rely on learning mechanisms in order to evaluate the potentiality of scanned spectrum bands and observe the strategies of other secondary users. A natural framework for studying such algorithms is game theory.

Doctorant.e: Iellamo Stefano