Towards Autonomous Management of Large-Scale Wireless Sensor Networks Utilizing Multi-Parent Recursive Area Hierarchies
Cree, Johnathan Vee
MetadataShow full item record
Large scale, duty-cycled, wireless sensor networks distributed across a large geographic area have been proposed for applications ranging from anomaly detection to vehicle tracking. In order to support the requirements of these applications an efficient and effective network management method that autonomously configures and maintains the network is required. When selecting a network management method it is important to consider both the direct and indirect costs associated with the different solution. An indirect cost that can be significant in duty-cycled networks is the overhead associated with local synchronization for communication. Further, an effective solution needs to recognize that in-network data aggregation and analysis presents significant benefits to wireless sensor networks and should configure the network to provide inherent benefits when said higher level functions utilize the structure as a paradigm. NOA, the proposed network management protocol, provides a multi-parent hierarchical logical structure for a network. NOA utilizes the multi-parent structure to reduce the cost of local synchronization as well as hierarchy creation and maintenance. A set of models has been designed and presented in order to compare the construction and data aggregation cost of NOA with other protocols. The ns-3 simulator was extended to provide a simulation environment for NOA and hierarchical beaconing and simulation data is presented to further verify the reduced management cost. The simulator also confirms that the reduced management costs are significant enough to extend the lifetime of a network configured with NOA compared to a network configured using hierarchical beaconing. The multi-parent structure provided by NOA can also provide higher level functions with benefits such as, but not limited to: removing network divisions that are encountered in single-parent hierarchies, data comparison between a device and its neighbors at a common grandparent, and redundancies for communication paths as well as in-network data aggregation, analysis and storage.