Event-Driven Data Collection Protocol Design For Wireless Sensor Networks

The development of wireless sensor networks (WSNs) has become a driving forceof improving sensing technologies. WSNs bridge the gap between physical and virtualworld, and then signifcantly increase the capability of gathering the information. Theapplications of WSNs are based on the sensing data collection, i.e., the process of col-lecting sensing data from sensor nodes to data sinks. Hence, designing energy efcientdata collection protocols for diferent applications is one of the most challenging andimportant issues in the research of WSNs.Based on the understanding and analysis of characteristics of WSNs and difcul-ties in designing data collection protocols, this thesis studied event-driven data collec-tion strategies in diferent network scenarios. These scenarios include:1) the wirelesssensor network with multiple sinks;2) the wireless sensor network with mobile sinks.According to distinctive features of each scenario, this thesis made following contribu-tions:1) For the wireless sensor network with multiple sinks, the strategy proposed inthis thesis selects several sensor nodes to represent the entire event area by exploitingthe spatial correlation among sensing data. The strategy intends to guarantee the accu-racy when rebuilding the event at sinks. The strategy jointly considers the selection ofrepresentatives and the selection of destination sink. This problem is frstly modeledas a Binary Integer Linear Programming with the optimization minimizing the ener-gy consumption of the data collection process. This is proved to be an NP-completeproblem. Due to the high computation complexity, two heuristic algorithms, correla-tion frst algorithm and distance frst algorithm, are proposed for approximation. Inthe proposed strategy, the collection method of spatially correlated data is extended from one sink to multiple sinks. Without sacrifcing the accuracy of collected data, thestrategy distributes the trafc load to diferent sinks, and largely decreases the numberof sensor nodes which take part in the data transmission for the collection. Thus, thestrategy can greatly improve the energy efciency of the data collection process.2) In the wireless sensor network with mobile sinks, this thesis focuses on theproblem of building connection between data sources and mobile sinks. Solving theproblem will lay the foundation for the transmission of event data. This thesis frstlyanalyzes the performance of several representative location service protocols under theevent-driven scenario. Based on the analysis, we fnd that these protocols can not ap-plyto theevent-drivenscenario directly. Then, weproposedthetrail-basedsearch(TS)strategy to build the connection between data sources and mobile sinks. TS strategyis based on Expanding Ring Search strategy, but the searching efciency is signif-cantly promoted because of exploiting the trail information. On the basis of proposedstrategy, we further discuss several features of the trail, including the trail length, theconnectivity of the trail, trail’s impact on the searching probability, with the purposeof providing supports in improving mechanisms and optimizing parameters.