Application-Aware Resource Management in Wireless and Visual Sensor Networks

Wireless Sensor Networks continue to gain tremendous popularity, as evidenced by the increasing number of applications for these networks. The limiting factors of the sensor nodes, such as their finite energy supplies and their moderate processing abilities, as well as the unreliable wireless medium restrict the performance of wireless sensor networks. In this dissertation, we explore application-aware techniques for managing the limited resources available to wireless sensor networks, focusing on ways to maximize the network lifetime for clustered sensor networks and visual sensor networks. We begin by focusing on a hierarchical cluster-based sensor network architecture, which reduces the network’s need for resources through data aggregation and a reduction of the communication overhead. In order to provide balanced energy consumption in clustered sensor networks, we introduce novel methods for network organization based on unequal size clusters, as opposed to the generally accepted equal-size clustering approach. In sensor networks that are constrained by coverage-preserving requirements we explore the use of application-aware (i.e., coverage-aware) metrics for cluster head, router, and active node role assignment. We believe that the real potential of wireless sensor networks lies in their integration with various sensing technologies and research disciplines, which will trigger widespread use of sensor networks in a number of specific applications. Thus, we adopt these ideas of application-aware resource allocation specifically for visual sensor networks. The higher demands for the network’s resources, the tighter QoS requirements, and the unique camera sensing capabilities are the main reasons why these types of networks are different and more challenging than existing (“traditional”) sensor networks. We examine the impact of camera-specific sensing characteristics on the network’s behavior in the case when a routing pro-