« Medium Access Protocol (MAC) Design for Multi-hop Wireless Ad hoc and Sensor Networks »
Qui ?
La thèse sera soutenue par Afef SAYADI, doctorant au sein du département LOR de Télécom SudParis.
Quand ?
Mercredi 16 janvier 2013 à 14h00.
Où ?
Télécom SudParis, 9 rue Charles Fourier, 91000 Evry, salle A007.
Jury
Rapporteurs :
M. Hartmut KONIG, Professeur à l’Université BTU Cottbus
M. Sidi Mohammed SENOUCI, Professeur à l’Université de Bourgogne
Examinateurs :
Mme Maryline LAURENT, Professeure à Télécom SudParis
M. Bachar WAHBI, Docteur à Montimage
M. Anis LAOUITI, Maître de Conférence à Télécom SudParis
Mme Ana CAVALLI, Professeure à Télécom SudParis (Directrice de thèse)
Résumé
« Wireless multi-hop ad hoc and sensor networks provide a promising solution to ensure ubiquitous connectivity for the Future Internet. Good network connectivity requires designing a reliable Medium Access Control (MAC) protocol, which is a challenging task in the ad hoc and sensor environments. The broadcast and shared nature of the wireless channel renders the bandwidth resources limited and expose the transmissions to relatively high collisions and loss rates. The necessity to provide guaranteed Quality of Service (QoS) to the upper layers triggered the design of conflict-free MAC protocols. The TDMA synchronization constraint is basically behind the rush of MAC protocol design based on a fixed frame size. This design shows inflexibility towards network variations and creates a network dimensioning issue that leads to a famine risk in case the network is under-dimensioned, and to a waste of resources, otherwise. Moreover, the alternative dynamic protocols provide more adaptive solutions to network topology variations at the expense of a fair access to the channel. Alongside with the efficient channel usage and the fair medium access, reducing the energy consumption represents another challenge for ad hoc and sensor networks. Solutions like node activity scheduling tend to increase the network lifetime while fulfilling the application requirements in terms of throughput and delay, for instance.
Our contributions, named OSTR and S-OSTR, address the shortcomings of the medium access control protocol design in the challenging environment of wireless multi-hop ad hoc and sensor networks, respectively.
For OSTR the idea consists in adopting a dynamic TDMA frame size that increases slot-by-slot according to the nodes arrival/departure to/from the network, and aiming to achieve a minimum frame size. For this end, OSTR couples three major attributes: (1) performing slot-by-slot frame size increase, (2) providing a spatial reuse scheme that favors the reuse of the same slot if possible, (3) and ensuring an on-demand frame size increase only according to the node requirements in terms of throughput. To tackle different frame sizes co-existence in the network, OSTR brings a cooperative solution that consists in fixing an appointment, a date when the frame size in the network is increased.
Concerning S-OSTR, it is an amendment of OSTR for wireless sensor networks. It brings the idea of a dynamic active period, since it deploys a dynamic frame size that is built slot-by-slot according to nodes arrival to the network. S-OSTR enforces the slot-by-slot frame size increase by a node activity scheduling to prolong the inactivity period in the network, and hence prolong the overall network lifetime for wireless sensor networks.
Our contributions are both based on the new dynamic TDMA frame size increase that consists in increasing the frame size slot-by-slot aiming to achieve a shorter frame size, and hence improve the channel utilization, and reduce the energy consumption. The performance analysis of OSTR and S-OSTR shows that they present good potentials to support QoS requirements, to provide energy-efficiency, to ensure fair medium access, to accommodate network topology changes and finally, to enhance robustness against scalability. The impact of this new TDMA frame size increase technique on the medium access control protocol performance is highlighted through multiple simulations of OSTR and S-OSTR. Multiple comparative studies are also handled to point out the effectiveness of this new technique and the soundness of our contributions. »