Towards the Internet of Water: Using Graph Databases for Hydrological Analysis on the Flemish River System

Abstract

The "Internet of Water" project will deploy 2500 sensors along the Flemish river system, in Belgium. These sensors will be part of a monitoring system. This will produce an enormous amount of data, over which prediction and analysis tasks can be performed. For representing, storing, and querying river data, relational databases are normally used. However, this choice introduces an "impedance mismatch" between the conceptual representation (typically a graph) and the storage model (relational tables). To solve this problem, this paper proposes to use graph databases. The Flemish river system is presented as a use case and the Neo4j graph database and its high-level query language, Cypher, are used for storing and querying the data, respectively. A relational alternative is implemented over the PostgreSQL database. A collection of representative queries of interest for hydrologists is defined over both database implementations.