- مبلغ: ۸۶,۰۰۰ تومان
- مبلغ: ۹۱,۰۰۰ تومان
Localization of unknown nodes in wireless sensor networks, especially for new coming nodes, is an important area and attracts considerable research interests because many applications need to locate the source of incoming measurements as precise as possible. In this paper, in order to estimate the geographic locations of nodes in the wireless sensor networks where most sensors are without an effective self-positioning functionality, a new graph embedding method is presented based on polynomial mapping. The algorithm is used to compute an explicit subspace mapping function between the signal space and the physical space by a small amount of labeled data and a large amount of unlabeled data. To alleviate the inaccurate measurement in the complicated environment and obtain the high dimensional localization data, we view the wireless sensor nodes as a group of distributed devices and use the geodesic distance to measure the dissimilarity between every two sensor nodes. Then employing the polynomial mapping algorithm, the relative locations of sensor nodes are determined and aligned to physical locations by using coordinate transformation with sufficient anchors. In addition, the physical location of a new coming unknown node is easily obtained by the sparse preserving ability of the polynomial embedding manifold. At last, compared with several existing approaches, the performances of the presented algorithm are analyzed under various network topology, communication range and signal noise. The simulation results show the high efficiency of the proposed algorithm in terms of location estimation error.
We have presented a graph embedding method for the location estimation problem in WSNs based on measured pair-wise distance. We view the sensors in the network as independently distributed devices and choose a suitable heat kernel function to measure the similarity between each pair of sensor nodes. Then we propose a novel localization algorithm under the manifold learning processes with an explicit nonlinear mapping based on the assumption that there exists a polynomial mapping between the localization data space and their low dimensional representations. For computing the geodesic distance graph, a suitable datum nodes m and polynomial degree q are selected in a right way. And then, the polynomial mapping method is presented and the relative locations are estimated by GEPM method. Finally, the physical coordinates of the sensor nodes can be obtained by the affine coordinate transformation. Experimental results confirm the promising performance of the proposed GEPM. Our future work will focus on the explicit nonlinear mapping of manifold learning localization methods for the target detection and tracking in the wireless sensor networks.