Research Of Data Interpolation Algorithm For Sensor Networks

The wireless sensor network has a broad application in the field of military and civilian, so it is an active area of research at present. Due to the low-cost property of the sensor nodes, the time sequence data produced by the sensor nodes contains a large amount of error data which needs to be regulated. Meanwhile, there are some invisible areas in the monitoring because of the limited sensor nodes which are deployed in the relatively bad deployment environment. In order to solve those problems, we need to fill up the data with interpolation. This paper studies how to solve the issues of data regulation and spatial interpolation in the wireless sensor network using Kriging Interpolation and Natural Neighbor Interpolation. The major jobs are as follows:Based on the summarizations of related knowledge in the wireless sensor and associated interpolation algorithms, two interpolation algorithms are proposed.(1) A grid_based fast Kriging Interpolation is presented. Firstly, it divides the region which is waiting for interpolating into many grids irregularly. Then it confirms the k-hops neighbor nodes of the forecast points and makes the neighbor nodes to be the referenced points assembled for Kriging interpolation. Finally, it can get the Kriging coefficients which can calculate the values of the predicted point. Compared with the general meshing method, the irregular meshing method can fully adapt to the irregular deployment of the sensor nodes. At the same time, by using the irregular meshing to search for the referenced points, it can reduce the complexity of the searching. The speed of Kriging algorithm’s execution is accelerated and the network energy is saved in a large part. What’s more, according to the further study of the construction of environment map, we introduce a method which is called gray environment map construction to construct the environment map. The method is based on the grids. This method greatly enhances the visualization of environment data sets, enabling us to get the environment at a glance.(2) The Natural Neighbor Interpolation based on Voronoi diagram is proposed. Instead of building a global Voronoi diagram, this algorithm establishes a local voronoi diagram. Firstly, it finds the related neighboring nodes of the node that is waiting for being interpolated by using the neighborhood query. Then construct the local Delaunay triangle net by using the neighboring nodes. Based on the local Delaunay triangle net, we can generate the local Voronoi diagram easily. Then update the local Voronoi diagram according to considering the awaiting interpolation node as a virtual node. Finally, we can predict the value of the awaiting interpolation node by calculating every neighboring node’s Voronoi area. Constructing a local Voronoi diagram saves time and space largely. Proved by the simulation experiments, the results of Natural Neighbor Interpolation Algorithm are closer to the real values compared to IDW.