LSD2H: A Novel Storage Method of Linked Sensor Data Based on HBase

With the development of sensor network, the number of sensors are increasing now. In order to manage sensor data, W3C has proposed semantic sensor web which unifies sensor network data and produces massive RDF dataset. Linked Sensor Data is a normal dataset of semantic sensor web and links into Linked Open Data based on SSN ontology. HBase is a kind of distributed database and suits for the management of Linked Sensor Data applicably. So we propose LSD2H which is a storage method of Linked Sensor Data by use of HBase. LSD2H architecture consists of SDS, ODS, SD2H, OD2H and SCQuery. SDS is a table in HBase for storing sensor data. ODS is a table in HBase for storing observation data. SD2H maps Linked Sensor Data to SDS, while OD2H maps Linked Sensor Data to ODS by MapReduce. SCQuery based on tree pattern and recursion algorithm is a query method about SDS and ODS. In order to deduce storage space and improve query performance, we analyze the compression algorithms of LZO, GZIP, Snappy and no compression, which has proved that LZO selected by using storage of data is benefit for LSD2H in final.