Resource estimation and control in wireless networks
Elizabeth Belding, Irfan Sheriff · 2008
The capacity of wireless networks is fundamentally limited due to a fixed frequency spectrum and the shared nature of the resources. The increased usage of IEEE 802.11 based wireless mesh networks as metro backhauls as well as for indoor connectivity and the growing popularity of real time applications, such as VoIP and video, presents a challenging resource management problem due to the limited capacity of wireless networks. IEEE 802.11 networks fail to support high traffic volumes [42]. Hence an unrestrained increase of user traffic can lead to degraded network performance and at times can cause complete network breakdown [42]. This can severely restrict the usefulness of mesh networks as last mile infrastructure for ubiquitous wireless connectivity. In this PhD dissertation, we address the problem of resource management in wireless networks to enable the support of media-rich applications with constrained QoS requirements. We believe that resource management is a critical component towards enabling ubiquitous multimedia support on wireless networks in cities, homes, offices, schools and university campuses. Our contributions can be broadly categorized into three main areas as follows. First, to assess the capability of wireless mesh networks, we perform experimental evaluations on deployed networks [75, 87]. We show that VoIP and video support on current 802.11 multihop networks is severely constrained. Additionally, we demonstrate the drawbacks associated with the current design of routing in mesh networks. Next, to predict and control the usage of available wireless resources, we provide a measurement driven framework for admission control on multihop wireless networks [82, 81]. Our work provides a decision making framework to predict the feasibility of admitting new voice calls on a network. Further, we improve upon the measurement based prediction framework to predict link interference on an indoor wireless testbed. This enables selection and control of link sets for packet scheduling on a TDM wireless network. Finally, to efficiently utilize the available resources in multihop wireless networks, we build an indoor multi-radio wireless network, the UCSB MeshNet, and propose channel assignment and multipath selection schemes that can significantly increase the network utilization to enable the support of multimedia applications [76, 83]. In addition to providing a framework to manage the restricted wireless resources, our work shows the feasibility of a distributed framework to make global predictions based on local inferences about the wireless conditions. This makes a strong case for measurement-based protocol design for future wireless networks.