Data Collection Algorithm Design In Mobile Wireless Sensor Networks

Mobile Wireless Sensor Networks provide low cost, short delay, high throughputsensing coverage for large areas by exploiting the mobility of the Wireless SensorNetworks. However, thespatial-temporal heterogeneous distribution of the mobiles willcause long delay and poor coverage of data collection from the sensors, and will alsoinfluence the performance of the routing algorithms. How to provide efficient datacollection is a major problem in mobile WSN.In this thesis, we first define the data delivery efficiency as the probability that dataof the whole network can be delivered to the base-station within a deadline. We furtherimprove such delivery efficiency through two procedures, namely, the data collectionand the data routing. In the data collection procedure, an energy efficient collaborativedata collection algorithm is proposed to minimize energy consumption of the staticsensors with respect to effective data collection. In the data routing procedure, we takethe network capacity constraint into consideration, and propose a density adaptive delaytolerant routing to fully utilize the limited capacity and improve data delivery ratiowithin deadline.Simulations on Beijing Taxi Trajectory Dataset, which includes GPS records of27,000taxis in Beijing in May,2009, are deployed to further verify the performance ofthese algorithms.Simulations show the data collection algorithm results in a50%reduction of energy consumption of the static sensors under the constraint of the sameeffective data collection ratio, comparing with traditional data collection schemes. Andthe density adaptive routing algorithm brings a delivery ratio improvement of20%comparing with other traditional DTN routing taking consideration of channel capacityconstraint.Algorithms proposed are measured under the metric of delivery efficiency. Suchmetric fully manifests thenetwork performance in real applications. Comparing withtraditional methods, these algorithms take distribution heterogeneity and capacityconstraint into consideration, and improve the data collection and coverage performancein large scale Mobile WSN. With distributed nature and low computational complexity,these algorithms can be easily deployed in Mobile WSN applications.