Study On The Target Localization Algorithms Based On Wireless Sensor Networks

In the recent decades,wireless sensor networks(WSN)have been applied in a wild rang of fields,due to their flexible structure and strong monitoring capability.Target localization is one of the most key technologies for WSN to realize their applications and functionalities.Therefore,it is of great theoretical significance and practical value to study localization methods for WSN.Typically,the process of wireless source localization involves a number of distributed anchored sensor nodes with known locations.The sensors first obtain the positioning measurements from the source signal received and then send the measurements to a data processing center.By cooperatively using these measured data,a series of positioning equations are constructed in the data processing center to estimate the location of the target.Commonly used measurements for positioning task include time of arrival(TOA),angle of arrival(AOA),time difference of arrival(TDOA),received signal strength(RSS)and so on.However,the problem is that these measurements are highly nonlinear and nonconvex with respect to the source position,which makes it a nontrivial task to directly solve the localization problem.Besides,in practical applications,source localization is seriously affected by non-line of sight(NLOS)propagation.In NLOS environments,the TOA positioning measurements abstracted by the sensor nodes from the received source signal often contain large and positive NLOS errors,which seriously reduce the localization accuracy.Hence,how to mitigate or eliminate the influence of the NLOS error is another challenge in source localization.Aim to the above problems,this dissertation explores and studies the source localization technology under different situations,and develops some efficient solutions.The main work of this dissertation includes the following aspects: 1.Basing on the two-way message exchange mechanism,a bi-iterative method is proposed for joint synchronization and localization problem of an unknown node in WSN.Most traditional approaches always simultaneously estimate the source node location and clock parameters,ignoring the fact that the original problem is linear and nonlinear with respect to the clock parameters and the source position parameters,respectively,which means solving the clock parameters separately can yield the closed-form solution.In order to fully make use of this characteristic,in the proposed method,the location and timings of the unknown source node are computed alternately by decomposing the original problem into two closely related sub-problems respect to the source node position and clock parameters,respectively.It can not only reduce the computational complexity and but also enhance the convergent speed.Convergence behavior of the proposed method is theoretically proved.Based upon the first-order perturbation analysis,the performance of the proposed algorithm is further analyzed.Theoretically analysis and simulation results indicate that the accuracy of the proposed method can approximately reach the Cramer-Rao Lower Bound(CRLB)under the mild noise condition.Compared with some previous methods,the proposed method is more computationally efficient,and takes fewer anchor nodes and less communication overhead,which is attractive to reduce the energy consumption for WSN.2.As we all known,TOA localization technology requires synchronization between the target and the sensor nodes,which always makes it failure to locate a non-cooperative target.To overcome this problem,the localization method based on the pseudo-range information is studied.As the signal transmit time of the non-cooperative target is not a known priori,the range measurements calculated from the measured TOA is actually equal to the real distance plus an unknown rang bias caused by the unknown transmit time.So the rang measurement is called pseudo-range.We first introduce two processing ways for pseudorange-based localization: TDOA processing and joint estimation processing.Then the nonlinearity of the two processing ways is analyzed and compared by using differential geometry curvature measures.The results shows that the joint estimation processing performs better than the TDOA processing in practice.Basing upon this,a joint estimation algorithm via constrained weighted least squares(CWLS)is proposed.In the proposed method,the original nonlinear positioning equations are converted into pseudo-linear ones by introducing an intermediate variable.Then,utilizing the relationship between the unknowns and the intermediate variable as a constraint condition,a CWLS optimization problem is formulated.Finally,by solving the problem with the technique of lagrange multipliers,the closed-form solution is obtained.The performance of the proposed method is theoretically analyzed.The experimental results demonstrate the effectiveness of the new algorithm.3.A new Two-Step Semidefinite Programming(TS-SDP)algorithm is developed to deal with the non-linear and non-convex problem of acoustic energy-based localization.The proposed algorithm first transforms the nonlinear localization problem into an approximate weighted least squares estimation problem of the unknown source location and signal transmit power,which is then solved in two steps.First,the signal transmit power is eliminated from the cost function by expressing it as a function of the source position in least square sense.In the second step,the WLS formulation is converted into a semidefinite programming(SDP)optimization problem by using a new convex relaxation technique.The tightness of the TSSDP algorithm is theoretically proved.Due to the approximate error to the original problem is smaller than the conventional methods,the TS-SDP algorithm has a better localization performance especially when the measurement error is relatively large.4.In non-line of sight(NLOS)conditions,the measurements obtained by sensor nodes always contain larger NLOS errors.The localization accuracy can be severely degraded if directly using the NLOS-corrupted measurements.To overcome this problem,we regard the NLOS measurements as outliers,and develop a novel localization algorithm named Soft Decision Direction Optimization(SDDO).In the SDDO algorithm,the cost function is formulated based upon the fact that the LOS measurements is consistent with position of the target.In order to effectively optimize the cost function,the deterministic annealing strategy is used.In the process of optimizing the cost function,as the temperature drops,the SDDO algorithm can automatically and gradually reduce the weighted value of the NLOS measurements to mitigate or even eliminate the influence of NLOS measurement on localization performance,which can dramatically improve the positioning accuracy.Simulation results indicate that the performance of the SDDO algorithm is close to localization performance when only using LOS measurements.