Green Networking : Apport de la Radio Cognitive

Abstract

Green Networking is a recent concept that refers to the processes used to optimize a network to make it more energy efficient. It is able to overcome the conflicts between transmission power and energy saving by providing an automatic and adaptive management of the radio parameters based on the needs. In this context, cognitive radio services are needed. However, the selection of the best available spectrum band to meet the QoS requirements of secondary users, while respecting the current regulatory context, is considered a major challenge. In this thesis, we propose three contributions: the first one is based on reinforcement learning for energy consumption minimization. The second is based on two bio-inspired approaches, namely: the flower pollination algorithm and the Cuckoo search for transmission parameter adaptation. The third is based on the TOPSIS decision method for the selection of the best available spectrum band. Through the three proposed approaches, we seek to reconfigure and adapt the parameters of the cognitive radio during transmission according to the user's application needs while ensuring better energy efficiency. The results obtained through a series of tests and simulations demonstrate a clear superiority of our proposals in terms of quality of service and energy efficiency