Combining geo‐referencing and network coding for distributed large‐scale information management

Abstract

The widespread and ubiquitous availability of Internet access enables the collective sharing of huge amount of data generated by heterogeneous sources. For example, the information, which will be exchanged among entities (sensors, people, and services) of future smart cities to enhance the security and lifestyle of their citizens, poses the challenging question of how this information can be efficiently and effectively maintained across the city. In this article, we propose a decentralized approach, based on the distributed geographic table (DGT) overlay scheme, which exploits geo-referenced information about nodes to achieve efficient data management. After recalling DGT main concepts, we illustrate the possible node types and how information can be published and retrieved within the network. To cope with the unavoidable node failures and disconnections, our approach leverages upon randomized network coding to increase the robustness of publish/retrieval operations. Evaluation is carried out through an extensive simulation analysis for a realistic urban scenario using the metrics of efficiency in data publication/search, resource availability, and storage occupancy requirements. Results show the approach effectiveness for large-scale sharing of geo-referenced information and tradeoffs between redundancy overhead and resource availability. A few results obtained with a preliminary DGT implementation are also presented in the paper

High‐performance computing: to boldly go where no human has gone before

“…The manuscript by Picone et al [151] proposes a decentralized peer-to-peer approach based on the distributed geographic table (DGT) overlay scheme, which exploits geo-referenced information 3160 EDITORIAL about nodes to achieve efficient data management. Geo-referencing enables searching for data fragments in the most probable direction.…”