Table 1: Features of the surveyed infrastructure abstractions.

Abstraction Data access method Technology interoperability Ontology Service discovery Metadata Processing

OGC SWE Publish/Subscribe, queried streams and historical values, events, and alerts A set of XML specifications and interfaces for sensor interoperability Ontology not mentioned, but SensorML describes uniform data structure and units, and O&M describes observation structure SensorML models and XML schema can be mined for SensorML processes (includes connected sensors and actuators); SPS can be used to test asset availability for a task SensorML contains process metadata and O&M contain observation metadata Through SensorML processes and process chains; SAS for creating notifications from pattern matching data

Bouillet et al. [30] Stream, but the accessing method is not described Through the input semantics that data sources must adapt OWL ontologies Data sources and PE outputs that match to a PE input can be discovered Not mentioned PEs can be interconnected; an algorithm to combine PEs into an application

GNS Query with streams and historical data Abstracted technologies are described as virtual sensors No ontology, but the virtual sensor describes the data structure freely Virtual sensor can be discovered with their metadata descriptions Each virtual sensor holds metadata descriptions SQL like processing on the virtual sensors; virtual sensors can be sources to other virtual sensors

SenseWeb Query with streams and historical data Uniform API, with data and time abstraction Not used, but the API homogenizes data format, but the method is not described For sensors, according to sensor type or location, and for transformers Not mentioned Transformers can convert units, aggregate, fuse, and visualize data

SensorMap Query with streams and historical data; location-based queries possible Through SenseWeb DataHub A lack of standard ontology discussed Services can be discovered from GeoDB meta-data GeoDB holds meta-data Aggregate geographically close sensors and display on a map

Lamses Events for context-awareness and queries for data retrieval Describes a common interface for attaching WSNs into it Not mentioned, but uses XML for homogenizing data Not mentioned An XML based meta-information database for complementary hardware information Context-aware event and data processing; internally integrates data for context-aware events

SeNsIM Query for real-time data and events Wrappers connect technologies to a mediator and data is unified in an XML Not mentioned, but an XML model is used for the data unifying Wrappers discover sensors from abstracted technologies Not mentioned. Not mentioned clearly

Smart-M3 Publish/Subscribe and query/insert data KPs can communicate through a SIB in a smart space without temporal connections Any ontology can be used and there is an Ontology API for each used ontology KPs can find different services through a SIB, if the semantics of their data match Not mentioned Not supported/possible