Font Size: a A A

Study On Parameter Estimating Of Location And Tracking In Wireless Sensor Networks

Posted on:2017-04-04Degree:DoctorType:Dissertation
Country:ChinaCandidate:M Q ZhuFull Text:PDF
GTID:1108330482487050Subject:Electronic Science and Technology
Abstract/Summary:PDF Full Text Request
As a core part of the Internet of Things technology, Wireless Sensor Network (WSN), which has the advantages of flexible networking and deployment, small volume, is a highly concerned field in both academia and industry. WSN can be used in a wide range of areas such as national defense, target tracking for battlefield, anti-terrorist, rescue, environmental monitoring, medical care, space exploration, traffic management, agriculture, and so on. However, it is confronted with many challenges due to diversification of specific business needs and complex working environments.In research and applications of WSN, sensor node localization and target tracking are the most fundamental problems. Compared to existing localization and tracking mechanisms, the main work and major contributions of this dissertation includes the following four aspects.1. To deal with problems existing in localization linearity error and Received Signal Strength Indication(RSSI) fitting, wireless channel fading model for indoor environments is studied, then a compound indoor-locating estimation algorithm based on limited iterative filter and scaled Unscented Kalman filter is proposed. The deterministic sampling filter and information fusion theorem are combined to construct localization parameter vector model. Through which raw data was calibrated, it converted the RSSI localization problem into an optimization problem of nonlinear equations. It can estimate the target position and the channel attenuation parameter simultaneously and introduce the noise statistic estimator to restrain the filtering divergence. The simulation experiment results show that compared to traditional approaches the Improved compound algorithm, which can always convege, is more accurate and robust in dealing with the problem of node localization in noisy indoor environments.2. For the inherent computational complexity of Particle Filter that affects the tracking precision, a time delay difference estimation algorithm based on improved cubature particle filter is proposed. Based on non-linear sub-optimization estimation theorem and target tracking theory, time different tracking parameter vector model is constructed. To overcome the phenomenon of particle degeneracy, the Cubature Kalman Filter(CKF) and Gauss-Newton(G-N) rule is used to incorporate the most current observations and to provide more accurate importance density function for the Particle Filter by virtue of the measurement updated state variable. In the case of a small number of particles, the simulation results demonstrate that proposed algorithm can achieve satisfactory prediction and tracking perfromance in target tracking by using TDOA. Furthermore, compared to other methods the proposed method is faster, more accurate and less complex.3. Based on the analysis of special positioning performance requirements, node deployment pattern and parameters of the network model, an improved localization algorithm based on modified average hop distances is proposed for belt-type sensor networks of special forms. On the basis of cooperation localization theory and network connection property, the parameter estimation model is established. By changing the structure of data packets sent by anchor nodes with broadcasting, weighting the average hop distance estimates of reference anchor nodes to modify the average hop distance, using steepest descent algorithm to improve the node positioning accuracy. When this algorithm is applied to the belt-type sensor network, the experimental simulation validates its low complexity, high precision and speed.4. Considering intrusion detection and target tracking in specific regions of some application environments, a portable device node is designed and developed to built a localization and tracking prototype system which could be applied in WSN. Then, by combining with node high-precision filter estimation algorithm, the tracking experiment for mobile target in specific region is completed.This thesis is focused on the research of sensor networks in indoor single scene, outdoor hybrid environment pattern and even belt-type environments with special topology structure. The application of methods are expanded from using limited iterative filter to utilizing nonlinear information fusion theory, step by step. These methods are also applied to construct WSN localization and tracking systems, which offers many ideas for solving the problem of localizing and tracking the target node in different environments. Furthermore, how WSN network localization and tracking technologies are employed to solve practical problems is explored in this thesis.
Keywords/Search Tags:wireless sensor network, localization, tracking, nonlinear filtering, information fusion
PDF Full Text Request
Related items