A Visual Paradigm for IoT Solutions Development - Interoperability and Open-Source Solutions for the Internet of Things - page 28

Networks) platform, which de nes a virtual sensor concept and provides the means for

almost zero-programming WSN application development (Aberer 2007 ). Another example
is the work described in (Mosta

zur 2010 ), which includes a framework for modeling,
simulation and code generation of complete WSN applications. Also, (Ghica 2008 ) illus-
trates an environment for the design, modeling and development of integrated WSN
applications. The above list of development environments and tools is non exhaustive,
given that most WSN middleware platforms (e.g., those listed in (Chatzigiannakis 2007 ))
come with some sort of tooling support for sensor applications. However, similar to the case
of RFID oriented development environments, these tools are overly focused on WSN
aspects. Hence, they tend to deal with the low-level details of sensor applications (e.g.,
sensor nodes communication, sensor data collection, data ltering, and data fusion) rather

than with their higher level composition into IoT applications. Furthermore, they ignore in
several cases key aspects of IoT applications, such as virtual sensors and data streams, as
well as the need to model non-trivial sensors and contexts.
Recently, software engineering solutions for the integrated development of general
IoT applications have also emerged (e.g., (Cassou 2010 ; Patel 2011 )). These solutions
leverage semantic models that capture the main elements of IoT applications (Patel
2013 ), such as sensors, actuators, and techniques for abstracting and modeling context in
IoT applications. Note that context abstraction is an essential element of pervasive and
context-aware computing (Dey 2001 ), and therefore an indispensable part of any
development environment for pervasive computing applications (Dimakis 2008 ).
In addition to context abstraction, the separation of programming concerns has been also
identied as a key ingredient of integrated development environments for IoT appli-
cations (Patel 2013 ). The above efforts towards integrated development of IoT appli-
cations rely on proprietary models for sensors and context modeling, which is a set-back
to their wider adoption. Furthermore, they do not take advantage of recent efforts
towards accessing the low-level functionalities of the sensors through high level Web-
based interfaces (such as CoAP (Constrained Application Protocol) (Colitti 2011 )). The
use of web-based interfaces is a trend for most on-line IoT platforms, notably those
enabling streaming of IoT data in cloud computing infrastructures. Prominent examples
are the Xively ( https://xively.com/ ) and Thingspeak ( http://www.thingspeak.com )
platforms, which provide Web based interfaces for accessing IoT data feeds in the cloud.
Nevertheless, these platforms provide very simplistic non-interoperable models and
formats for sensors and context, while they do not provide common semantics for
interoperable representations of sensors. Therefore their associated visual development
capabilities are generally limited to very simple IoT applications.
One of the main bene

ts of the approach that is suggested in this paper is that it
relies on semantic models and ontologies under standardization (Serrano 2014 ), such as
the W3C Semantic Sensor Networks (SSN) (Taylor 2011 ), which provides the means
for abstracting/virtualizing virtually any sensor. In addition to general purpose
descriptions of sensors and observations, the SSN ontology provides the means for
modeling/abstracting the context of multi-sensor applications in order to facilitate
processes such as dynamic discovery of sensor data and metadata, as well as tasking
and programming in multi-sensor applications. In addition to its semantic power and
expressiveness, the W3C SSN ontology enables the use of web based technologies and
techniques for accessing and linking both sensor data and metadata, thereby giving rise