266
Internet of Things (IoT)
little effort. The major challenge for the IoT would be to help developers in designing and
developing interoperable inter-domain IoT applications. As devices are not interoperable
with each other due to their proprietary formats, they do not use vocabulary to describe
interoperable IoT data. One way to make them interoperable would be a common protocol
used by all devices. Another solution would be to work on the interoperability of this data,
since these devices are already deployed and data are already produced. Exploiting, combining, and enriching device data to build smarter interoperable applications is becoming
a real challenge. The growing linked open data encourages to share the data on the web,
including sensor data. To assist users or even machines in interpreting and combining the
sensor data, there is a real need to explicitly describe sensor measurements according to
the context, in a unified way and in a manner understandable by machines. For instance,
temperature measurement does not mean the same always and its meaning varies according to the context (room temperature, body temperature, water temperature, or external
temperature) and the machine will not infer the same knowledge (fever deduced with
body temperature, abnormal temperature for room temperature). We also need to deal
with implicit units (e.g., Fahrenheit, Celsius, Kelvin).
The second biggest challenge is combining domain-specific applications and data together
to create innovative cross-domain IoT applications. Existing applications are specific to one
domain such as smart home, smart health care, transportation, and smart garden. Examples
of innovative cross-domain IoT applications: (1) suggesting food according to the weather
forecast, (2) suggesting home remedies according to health measurements, and (3) suggesting safety equipment in a smart car according to the weather, etc. Future smart fridges will
enable to purchase groceries online. In case of RFID tags embedded in food, it will be easy
to recommend the menu for dinner or automatically order essential ingredients. If you are
an athlete, the smart fridge will recommend you the perfect diet in case of compatibility
with Apple Nike shoes. Finally, Google’s car could automatically take you to the grocery
store to grab the missing ingredients. Last but not the least, security issues should be considered when combining IoT data or designing IoT applications. For instance, health data
are more sensitive than weather data and need to be secured.
Further, according to Barnaghi et al. (2012), semantics is required at different levels in IoT;
it can be used to: (1) describe things and data, (2) reuse domain knowledge, (3) interpret IoT
data, (4) provide smarter applications, and (5) provide security (Figure 13.1).
The existing challenges for the semantics-enabled IoT can be summarized as follows:
Generating interoperable cross-domain semantic-based IoT applications. This process
should be flexible enough to be performed either on the cloud, constrained devices, or
machine-to-machine (M2M) gateways. M2M gateways mean that processing is done automatically, without requiring human intervention.
Interpreting sensor data and inferring new knowledge by reusing domain knowledge
expertise. Reusing domain knowledge (e.g., ontology) is highly recommended (Simperl
2009; Suárez-Figueroa 2010). The interoperability of domain knowledge enables building
of cross-domain expertise.
Securing IoT applications when designing these applications. This challenge should be
solved using the same approach as for the two previous challenges by combining security
knowledge expertise to find the most suitable security mechanisms to secure IoT applications.
To deal with the challenges, we exploit semantic web technologies (Berners-Lee et al.
2001) for several reasons. First, semantics enables an explicit description of the meaning of
sensor data in a structured way, so that machines could understand it. Second, it facilitates
interoperability for data integration since heterogeneous IoT data is converted according to
the same vocabulary. Third, semantic reasoning engines can be easily employed to deduce
Internet of Things (IoT)
little effort. The major challenge for the IoT would be to help developers in designing and
developing interoperable inter-domain IoT applications. As devices are not interoperable
with each other due to their proprietary formats, they do not use vocabulary to describe
interoperable IoT data. One way to make them interoperable would be a common protocol
used by all devices. Another solution would be to work on the interoperability of this data,
since these devices are already deployed and data are already produced. Exploiting, combining, and enriching device data to build smarter interoperable applications is becoming
a real challenge. The growing linked open data encourages to share the data on the web,
including sensor data. To assist users or even machines in interpreting and combining the
sensor data, there is a real need to explicitly describe sensor measurements according to
the context, in a unified way and in a manner understandable by machines. For instance,
temperature measurement does not mean the same always and its meaning varies according to the context (room temperature, body temperature, water temperature, or external
temperature) and the machine will not infer the same knowledge (fever deduced with
body temperature, abnormal temperature for room temperature). We also need to deal
with implicit units (e.g., Fahrenheit, Celsius, Kelvin).
The second biggest challenge is combining domain-specific applications and data together
to create innovative cross-domain IoT applications. Existing applications are specific to one
domain such as smart home, smart health care, transportation, and smart garden. Examples
of innovative cross-domain IoT applications: (1) suggesting food according to the weather
forecast, (2) suggesting home remedies according to health measurements, and (3) suggesting safety equipment in a smart car according to the weather, etc. Future smart fridges will
enable to purchase groceries online. In case of RFID tags embedded in food, it will be easy
to recommend the menu for dinner or automatically order essential ingredients. If you are
an athlete, the smart fridge will recommend you the perfect diet in case of compatibility
with Apple Nike shoes. Finally, Google’s car could automatically take you to the grocery
store to grab the missing ingredients. Last but not the least, security issues should be considered when combining IoT data or designing IoT applications. For instance, health data
are more sensitive than weather data and need to be secured.
Further, according to Barnaghi et al. (2012), semantics is required at different levels in IoT;
it can be used to: (1) describe things and data, (2) reuse domain knowledge, (3) interpret IoT
data, (4) provide smarter applications, and (5) provide security (Figure 13.1).
The existing challenges for the semantics-enabled IoT can be summarized as follows:
Generating interoperable cross-domain semantic-based IoT applications. This process
should be flexible enough to be performed either on the cloud, constrained devices, or
machine-to-machine (M2M) gateways. M2M gateways mean that processing is done automatically, without requiring human intervention.
Interpreting sensor data and inferring new knowledge by reusing domain knowledge
expertise. Reusing domain knowledge (e.g., ontology) is highly recommended (Simperl
2009; Suárez-Figueroa 2010). The interoperability of domain knowledge enables building
of cross-domain expertise.
Securing IoT applications when designing these applications. This challenge should be
solved using the same approach as for the two previous challenges by combining security
knowledge expertise to find the most suitable security mechanisms to secure IoT applications.
To deal with the challenges, we exploit semantic web technologies (Berners-Lee et al.
2001) for several reasons. First, semantics enables an explicit description of the meaning of
sensor data in a structured way, so that machines could understand it. Second, it facilitates
interoperability for data integration since heterogeneous IoT data is converted according to
the same vocabulary. Third, semantic reasoning engines can be easily employed to deduce