| Wireless Sensor Networks, consisting of sensors equiped with energy-limitedbatteries, is the basic components of the Internet of Things. Due to the seriousenergy constraint imposed on such a network, it is a great challenge to performgathering data efficiently. In this dissertation, we consider the data gathering anddata fusion problem in wireless sensor netwoks with an objective to prolong thenetwork lifetime. The following are the major works of the dissertation.1. A data gathering protocol called DEEC-MR is proposed based on clusteringand multilevel routing. In the protocol, a node according to its residual energy,independently makes its decision to compete for becoming a cluster head. Eachcluster head finds a parent node according to the distance between neighbor clusterhead and sink node, residual energy of neighbor cluster head and so on. Thus a nearoptimal mimum spanning tree rooted on sink node is constructed. Data gathered bya cluster is transmitted along the spanning tree to sink node in a multi-hop manner.Experimental results have shown that DEEC-MR significantly outperform the othertwo protocols(LEACH, PEGASIS) in terms of network lifetime. Moreover,DEEC-MR balances the energy consumption to avoid some node dying prematurely,so the reliability is higher than LEACH and PEGASIS.2. A feature-level data fusion algorithm called FLAKP is proposed, whichreduces the consumed energy at cluster head by Kalman filter and principalcomponent analysis. The simulation results show that the fusion algorithm reducesthe amount of data transferred to sink node effectively while preserving most of theenvironment information.3. A decision-level data fusion algorithm called DLAAKP is proposed basedon FLAKP, in which cluster head make decision by neural network. Only theabnormal output data of neural network will be sent to sink node. The simulationresults show that the data fusion algorithm can make right judgements anddecisions on the state of the environment. |
PDF Full Text Request