Research On Node Localization In Sensor Array Networks

Location information of sensor nodes in sensor array networks (SAN) is the basis for many applications. Node localization in SAN has long been considered as a challenging task owing to its severe constraints on cost and energy. This thesis mainly studies the localization algorithms and schemes for direction of arrival (DOA) based node localization so as to improve the localization accuracy and reduce the cost. The main contributions in this thesis are summarized as follows:1. This thesis develops a set of efficient closed-form localization algorithms derived from ef-fective auxiliary variables based method. First, this thesis formulates the self-localization problem as a linear least squares problem using auxiliary variables and then develop a new auxiliary variables based pseudo-linear estimator (AVPLE). By analyzing the estimation er-ror of the AVPLE algorithm, a bias compensated AVPLE (BCAVPLE) is presented to re-duce the estimation error. In the case of the accurate and inaccurate prior knowledge of the AOA noise variance, respectively, a novel BCAVPLE based weighted instrumental vari-able (BCAVPLE-WIV) estimator and a new AVPLE based WIV (AVPLE-WIV) estimator are developed to estimate the locations and orientations of unknown nodes. Also, an au-tonomous coordinate rotation (ACR) method is proposed to overcome the tangent instability of the proposed algorithms when the orientation of the unknown node is near π/2. Extensive simulation results demonstrate that the new algorithms achieve much higher localization ac-curacy than the triangulation method and also avoid local minima and divergence problems in iterative ML estimators.2. In order to overcome the tangent instability of the proposed algorithms when the orientation of the unknown node is near π/2, an auxiliary variables based total least square (AVTLS) es-timator is proposed to estimate the unknown nodes instead of ACR algorithm. In addition, to further improve the localization accuracy, an AVTLS based WIV (AVTLS-WIV) estimator is developed to achieve asymptotically unbiased estimation of the unknown nodes. Simulation results demonstrate that the proposed AVTLS and AVTLS-WIV estimator have higher lo-calization accuracy than the AVPLE. Moreover, their estimation performances remain stable even if the orientation angle of the unknown node varies.3. To improve the localization accuracy and reduce the cost, a novel positioning method using auxiliary sound sources (PASS) is proposed to localize sensor nodes. An auxiliary vari-ables based total least squares estimator (AVTLS)is proposed to estimate the location and orientation of each unknown node. For performance evaluation purpose, this thesis com-pares the localization accuracy, the cost and the node coverage of PASS with the traditional DV-Bearing, and exploit the effect of the number of auxiliary sound sources, beacons, and unknown nodes on the localization accuracy. Experimental results show that with the same number of beacons the localization accuracy and node coverage of PASS constantly surpass-es that of the DV-Bearing.