Efficient algorithms and protocols for stability management in mobile ad hoc networks

This dissertation addresses issues pertaining to mobile multi-hop radio networks, also called ad hoc networks, which play a critical role in places where a wired backbone is neither available nor economically feasible and thus deployment of such networks becomes necessary. The main contributions of this dissertation are in the areas of clustering, routing and topology management. The research goals are to make progress towards achieving stability management in mobile ad hoc networks. Since the movement of the nodes (such as users having wireless mobile devices) constantly changes the topology of the underlying network, it becomes necessary to identify certain nodes, called cluster-heads, which help to maintain the topology of the network.; Our Weighted Clustering Algorithm (WCA) takes into consideration several physical parameters, such as the number of nodes a clusterhead can handle ideally, transmission power, mobility, and battery power of the nodes. We optimized WCA using the genetic algorithmic approach and obtained improved performance. To manage essential network operations, such as routing, we define a new protocol called B-protocol based on the fact that efficient management of network resources can be obtained by deploying only a small subset of the network nodes, called B-nodes. Routing is achieved via links, called B-links, which are also a small subset of the total number of links in the network. Since a route is not guaranteed to last till the end of a session, we estimate the expected lifetime of a route based on the relative mobility of nodes and their mobility pattern so that alternative routes can be found ahead of time to avoid any possible communication disruption. The stability of the network is also maintained by applying load balancing to the nodes, which may have different processing and battery powers, in the system as much as possible. The execution time of the jobs in each node is minimized while the lifetime of the nodes are maximized.