Sensor Network Localization Based on Distance Reconstruction

This paper proposes a novel distance matrix reconstruction (DMR) algorithm from incomplete and inaccurate distance information for nodes localization in wireless sensor networks. By using DMR algorithm, we address an improvement for self-organizing isometric embedding (SIEMAP) based localization in which DMR is used instead of the original distance matrix. The core of DMR is finding an Euclidean distance matrix to approximate the incomplete and inaccurate original distance matrix, and DMR based SIEMAP algorithm will be more robust for node localization when distance measurements are incomplete and inaccurate. Simulation results demonstrate that DMR can improve the location estimation effectively when the original distance matrices are incomplete and inaccurate. The results can be also applicable to many localization methods, such as dwMDS, SDP, Steep-SDP and so forth. In conjunction with proposed SIEMAP algorithm, we illustrate that the DMR based localization is robust and accurate for incomplete and inaccurate distance measurements.