Low cost localization using Nyström extended locally linear embedding

In this article, we employ Nyström method with locally linear embedding (LLE) to approximate the node location and develop an iterative localization technique for accurate node localization. Initially, a group of three adjacent anchor nodes forms a reference structure and nodes in one hop neighborhood of this reference structure are localized using LLE. After this, remaining sensor nodes are localized using Nyström extended LLE. Experimental results show that the proposed Nyström extended LLE (NLLE) is able to localize the nodes accurately for different value of anchors, connectivity and varying size of the neighborhood. It is also observed that the NLLE is able to localize sensor nodes with an increased accuracy at low cost as compared to existing manifold learning techniques.