Energy-efficient information processing and routing in wireless sensor networks: Cross-layer optimization and tradeoffs

Wireless sensor networks have become an important technology to realize many applications, including both simple event/phenomena monitoring applications and heavy-duty data streaming applications. While many systems are being developed, we focus on two fundamental operations: information processing and information routing. These two operations are tightly related and must be performed in a collaborative fashion.; A major concern in designing and operating sensor networks is their energy-efficiency. Cross-layer optimization is widely accepted as an effective technique to ameliorate this concern. In this thesis, we investigate three specific problems in the emerging and important research area of applying cross-layer optimization in the context of collaborative information processing and routing.; The first research effort addresses collaborative data processing in a single hop cluster that behaves as a basic operating unit across the network. We investigate the assignment and scheduling of a set of real-time communicating tasks onto the cluster under a novel performance metric---to balance the energy cost of all nodes within the cluster. We explore the energy-latency tradeoffs with adjustable computation and communication speed using voltage scaling and rate adaptation. The proposed 3-phase heuristic achieves up to 10x lifetime improvements in simulated scenarios.; The second research effort considers the transportation of information to the base station over an existing routing substrate (i.e., a data gathering tree) within a user-specified latency constraint. We again explore the energy-latency tradeoffs through rate adaptation. By exploiting the dependency between communication links over the tree, we propose techniques that achieve up to 90% energy conservation.; The third research effort investigates the construction of a routing tree that minimizes the total energy costs of data compression and communication. Such an objective is novel and crucial for advanced computation-intensive applications where a balance between computation and communication energy is necessary. We utilize the concept of tunable compression that explores the tradeoffs between the compressing time and the output size. We show that the Minimal Steiner Tree is a practical solution with constant performance bound for grid deployment and acceptable performance for systems with arbitrary deployment.