Skip to Main Content
Most sensor networks require application-specific network-wide performance guarantees, suggesting the need for global and flexible network optimization. The dynamic and nonuniform local states of individual nodes in sensor networks complicate global optimization. Here, we present a cross-layer framework for optimizing global power consumption and balancing the load in sensor networks through greedy local decisions. Our framework enables each node to use its local and neighborhood state information to adapt its routing and MAC layer behavior. The framework employs a flexible cost function at the routing layer and adaptive duty cycles at the MAC layer in order to adapt a node's behavior to its local state. We identify three state aspects that impact energy consumption: 1) number of descendants in the routing tree, 2) radio duty cycle, and 3) role. We conduct experiments on a test-bed of 14 mica2 sensor nodes to compare the state representations and to evaluate the framework's energy benefits. The experiments show that the degree of load balancing increases for expanded state representations. The experiments also reveal that all state representations in our framework reduce global power consumption in the range of one-third for a time-driven monitoring network and in the range of one-fifth for an event-driven target tracking network.