IPML-ANP: An integrated polynomial manifold learning model and anchor node placement for wireless sensor node localization

In Wireless Sensor Network (WSN), the geographical location of the sensor nodes is determined with localization techniques. The conflict between anchor node location, localization accuracy, and coverage must be resolved. In this article, the novel localization approach based on the location selection strategy of static anchor nodes is proposed. It combines the Polynomial manifold learning (PML) model with the anchor node placement approach, which derives the topology map from the distribution map of WSN sensor nodes. An ultra-fast Top-k Closeness Centrality (UF-TKCC) algorithm is introduced to choose the placement of anchor nodes. Each node's proximity centrality value is calculated using UF-TKCC, and the results are sorted to choose the first anchor node. To find the remaining anchor nodes, the centrality of closeness values are traversed at equal intervals. The physical coordinates of the unknown nodes are estimated using the pair-wise geodesic distance. The inherent geometrical feature is maintained using these information coordinates, which are also employed to build the weight matrix. After that, the proposed model will compute the polynomial matrix and the symmetric positive-definite (SPD) matrix using the supplied polynomial parameters. Experimental outcomes show that the IPML-ANP approach is more efficient than the traditional approaches with respect to localization accuracy.