Since the development of first packet-switched networks as specific research projects or dedicated networks for simple data transfer between computers, networks and their interconnection have been evolving and exhibiting various technological and architectural advancements. Networks are becoming very complex and hard to control especially due to their heterogeneity and the abundance of their internal components. If most of the current applications are able to provide their intended services, it is thanks to control mechanisms. They have mainly two contradictory objectives which are maximizing the utilization of the network and also maximizing the user satisfaction. In our research, we have focused on studying wireless networks and we have followed their evolution. We aim mainly at taming the complexity of the studied system via a metric or a model to come up with a controllable efficient mechanism that improves the system performance.
Accordingly, we present in this dissertation our research results related to three wireless technologies: WiFi-based multihop wireless networks, WiMAX mesh networks and recently cognitive radio networks. We mention also our works on wired networks dealing with the interaction between control packets and data. Throughout this dissertation, we provide insights learned from our past and current research works and develop our perspectives.
Particularly, we show that it is gainful to deploy some specific techniques to control the transmission in both the MAC and the network layers in WiFi-based multihop wireless networks. In WiMAX mesh, we show how scheduling and routing should adapt to different burst profiles of WiMAX links, and we highlight the impact of bottlenecks. In cognitive radio networks, metrics should be able to capture the availability patterns and the traffic requirement. Especially, spectrum handoff can be guided efficiently by a delay requirement.