End-to-end Delay Analysis for Event-driven Wireless Sensor Network Applications

The end-to-end delay is one of the most critical and fundamental issues for wireless sensor networks. Many applications of sensor networks require an end-to-end delay guarantee for time sensitive data. However, the end-to-end delay is difficult to bound for event-driven sensor networks, where nodes generate and propagate data only when an event of interest occurs, thereby producing unpredictable traffic load. Meanwhile, the end-to-end delay is tightly banded with many other factors, e.g., energy and network capacity. In this dissertation proposal, we analyze the quantitative relation among the end-to-end delay, the probability of guaranteeing this delay, and network parameters for event-driven sensor networks. We consider two sub-delays of the end-to-end delay, i.e., detection delay and queuing delay, based on a two-phase model. This two-phase model includes an event observation phase and a data propagation phase. We use target tracking as the example of an event-driven wireless sensor network application. For the event observation phase, we present a target prediction and sleep scheduling scheme, called TPSS, to reduce the energy consumption, while satisfying the given delay constraint. Wireless sensor nodes are typically in the sleep state most of the time to prolong the network lifetime, but this will increase the detection delay. Proactive wake-up and sleep scheduling are commonly used approaches to solve this problem. The purpose of TPSS is to select the nodes to be sleep-scheduled and reduce their wake-up time as much as possible, so as to enhance the energy efficiency as well as satisfy the given detection delay constraint. First, we design a target prediction method based on kinematics rules and theory of probability. Then based on the prediction results, we design a novel sleep scheduling mechanism that reduces the number of awakened nodes and schedules their sleep patterns in an integrated manner for enhancing energy efficiency. We analyze the detection delay and the detection probability under TPSS, and conduct simulation-based experimental studies. Our simulation results show that TPSS achieves a better tradeoff between energy efficiency and tracking performance than existing works. For the data propagation phase, we leverage queuing theory, and analyze the delay from the point of view of network capacity. In a many-to-one data gathering network, the throughput of a node can be estimated based on its distance from the sink node. Thus, the expected waiting time of a packet in the queue, i.e., the queuing delay, can be estimated by approximating all the nodes that …