Kriging for Localized Spatial Interpolation in Sensor Networks

The presence of coverage holes can adversely affect the accurate representation of natural phenomena being monitored by a Wireless Sensor Network (WSN). Current WSN research aims at solving the coverage holes problem by deploying new nodes to maximize the coverage. In this work, we take a fundamentally different approach and argue that it is not always possible to maintain exhaustive coverage in large scale WSNs and hence coverage strategies based solely on the deployment of new nodes may fail. We suggest spatial interpolation as an alternative to node deployment and present Distributed Kriging (DISK), a localized method to interpolate a spatial phenomenon inside a coverage hole using available nodal data. We test the accuracy and cost of our scheme with extensive simulations and show that it is significantly more efficient than global interpolations.