Managing wireless mesh networks: a measurement-based approach
Prasant Mohapatra, Chen‐Nee Chuah, Dhruv Gupta · 2010
Network Management can be defined as the deployment, integration, and coordination of hardware, software, and human elements to monitor, analyze, evaluate, and control network resources, in order to meet the operational performance and Quality of Service requirements in a network [88]. It includes the ability to monitor and gather data from various distributed network entities, and utilize this information for the purpose of maintaining network performance. Network management includes performance management, fault management, and network configuration and accounting. Wireless Mesh Networks are multi-hop wireless networks that have become increasingly prevalent over the last few years. They are being applied to an increasing number of applications ranging from municipal and enterprise networks to wildlife monitoring. However, apart from the usual intricacies involved in a wireless network, wireless mesh networks involve data transfer over multiple wireless hops, which adds to the complexity of maintaining network performance. The increased deployment and popularity of wireless mesh networks calls for new and effective management schemes to be in place, for the efficient working of the network. As part of our research, we adapt a novel approach to designing and implementing network management schemes for wireless mesh networks. Our novel design involves measurement-based schemes that incorporate the current status of the mesh network into the decision making process. Such an approach provides significant advantages over the more prevalent model-based schemes. Our novel implementation involves using efficient monitoring techniques in order to measure the necessary parameters for our measurement-based schemes. We implement a variety of passive monitoring modules that help us minimize monitoring-related overheads. We focus on three core areas within network management as part of this dissertation. We first investigate the aspect of efficient monitoring in wireless mesh networks. This study is a crucial first step in our research, as it enables us to identify the impact of network monitoring overheads on end user traffic. The second aspect of network management that we study is that of resource control. Based on the inferences from our previous study, we implement passive monitoring modules in order to measure various network parameters. We implement an admission control scheme, based on end-to-end delay as the design metric, which helps us in regulating the amount of traffic in the network. We also implement an association control scheme that utilizes available bandwidth as the design metric. This scheme enables end users to make an intelligent access point selection decision. We demonstrate the advantages of these approaches in terms of achieving better network performance, and enabling effective resource control and provisioning. Lastly, we consider the problem of fault diagnosis in wireless networks, which also falls under the realm of network management. We design and implement a novel framework wherein unique signatures are maintained for different network faults. These signatures help us identify the occurrence of a particular fault in the network, and also enable us to distinguish faults from each other, or from legitimate network events.