Distributed Information Storage And Retrieval In 2D/3D Wireless Sensor Networks

Wireless sensor networks bear the task of intensive data collection through their large scales and dense deployments. However, the limited energy storage of a node heavily confines WSNs' ability of accomplishing their missions. Making effective use of the vast amount of data gathered by large-scale sensor networks requires scalable and energy-efficient data storage and data dissemination algorithms. The utility of a sensor network derives primarily from the data it gathers. Distributed in-network data-centric processing aims to reduce energy consumed for communication and establish a self-contained data storage, retrieval, aggregation, and query sensor system which focuses more on the data itself rather than the identities of the individual network nodes.In real-world scenarios, it is common that the sensor field is irregular, possibly containing obstacles/holes of arbitrary shape. We propose a convex partition based information storage and retrieval approach for general 2D sensor networks with possibly complicated topology and geometric shapes. We should first divide the network field to subareas with convex shapes, and use double rulings scheme inside each subarea. For general 3D sensor networks with possibly complicated topology and geometric shapes, we propose a geographic location free projective geometry based approach. A general 3D sensor network is mapped to a sphere based on the Ricci flow algorithm and the inverse of stereographic projection. For high genus 3D sensor networks, we propose a geographic location free surface segmentation based approach. We should first slice the high genus 3D surface,to form a genus-0 surface with only one boundary. Finally the surface is mapped to a planar disk. Without the knowledge of the geographic location and the distance bound, a query simply travels along a simple curve with the guaranteed success to retrieve aggregated data with one or different types across the network. We compare the performance of proposed algorithms and GHTs for both retrieval quality and load balancing. Extensive simulations and comparisons show the proposed algorithms with low cost and a balanced traffic load.