The Research Of Data Fusion And Target Location In Wireless Sensor Networks

As a novel information acquisition technology, WSN has a great potential for both military and civil application. The challenging issues in sensor networks are limited energy provision, communication bandwith and computing power. Several methods about data fusion and target location are investigated to improve the energy efficiency, to reduce the effect of channel disturbance and to save the cost of calculation. The main contributions can be summarized as follows.1. Sub-optimal decision fusion under fading channels in wireless sensor networks is proposed. Firstly a sub-optimal fusion rule CSC with less channel parameter is derived. Secondly sub-optimal fusion rule RCN without channel parameter are proposed that are shown to be the low and high signal-to-noise approximation. Error processing theory is applied to the fusion rules. Simulation results show that the proposed fusion rules exhibit better performance.2. Multi-hop decision fusion based on cluster routing is proposed. Parallel fusion model that involves one hop and two hop transmission is studied and corresponding likelihood fusion rule and sub-optimal rule are derived.3. Weighted data fusion method based on medium theory is proposed. The measure of medium truth scale is used in data fusion. The sensor's weight is defined according to the function of distance ratio. The method need not estimate the variance by storing the history data of the sensor node, so dynamic fusion can be realized in real-time when the measurements change fastly over time. Simulation results show that the algorithm exhibits better evaluation accuracy and good ability of anti-disturbance of channel.4. Nonlinear equation location algorithm based on acoustic energy is improved. Adding samples and searching the maximum values of energy are used to resolve the equation to reduce the case of no solution. The method is simple and spends less calculation time. Simulation results show that it has a good performance of estimation with a few nodes.5. The model of multiple-source location is created based on the quantized data of acoustic energy. The maximum-likelihood location estimator of multiple target is derived to reduce the the bit transmitted in condition of given estimation performance.