An Efficient Spatial Data Index Based On Gossip for Overlay Networks

Abstract

In large and highly dynamic distributed environments, the retrieval of spatial data entails some challenges such as network topology, indexing efficiency, and mapping methods for multidimensional spatial data. Conventional methods may not allow for the complex retrieval of multidimensional spatial data in an efficient manner. This paper proposes an overlay network for indexing spatial data based on gossip. The proposed network, referred to as the 2D-Ring, has a ring-like topology with a spatial data index. First, the paper introduces a protocol for constructing an overlay network for indexing spatial data and then presents a method for mapping the data onto the 2D-Ring by using a distributed quadtree and control points. The results of a comprehensive experiment for an evaluation of the performance of the 2D-Ring with respect to various features indicate that the 2D-Ring converges logarithmically to network size, has desirable small-world properties, and is robust across dynamic scenarios. The proposed solution can be applied not only to overlay networks but also to other types of networks requiring the sharing of distributed data sets.