Page 168 - Proceedings of the 2017 ITU Kaleidoscope

2017 ITU Kaleidoscope Academic Conference

Fig. 5. Scaled down Sequence Diagram Showing relationship
among the Publisher (Sensor), Subscriber (u), Sensing Broker
and the proposed Framework
new subscriber is available, the broker notifies the TCPD
system to deliver the data to the subscriber along with the
Fig. 4. High-level Implementation Architecture of the proposed trust level of the data.
System
pollution/environment sensors. The data collected from the 6. CONCLUSION
air pollution sensors are delivered to the IoT Cloud, hosting In this work, we argue that the traditional means of trust
the TCPD proposed in this paper. Thus, a mobile app for computation for entities does not necessary guarantee the
trusted air quality data monitoring can be developed on top trustworthiness of data that they generate. Hence, we
of this framework integrating data collected from low-cost propose a hybrid trust computational platform which is
environment sensors for temperature, humidity, CO, CO2 capable of assessing both data centric trust as well as
NO2, SO2, as well as compounds including benzene and lead traditional entity based trust. Further, we provide a model to
(VOCs), etc. The sensors’ readings will be transmitted via compute individual DTA and the main DTM by combining
either an Android or IOS app to the proposed system for numerical models with learning algorithms. Afterward, a
assessing and predicting the trust of the data before it is sent data trust prediction scheme based on collaborative filtering
to the IoT Cloud. Such data can then be visualized along with is proposed to find the data trust between trustors and data
its trust level by interested individuals, government, city sources who do not have prior encounters that avoids using
administrators etc. via a web application. data from malicious actors. Finally, a possible

For the above use case to profit from the proposed solution, implementation scenario is discussed based on a crowd
we have proposed a distributed publish-subscribe sensing use case. Similarly, our algorithm would be
architecture such as CoreDX distributed publish subscribe beneficial to filter out malicious data and data sources to
middleware [36] whereby an interested parties can subscribe maintain integrity and quality of the outcomes that any
via a broker to environmental data of interest in specific crowd sensing application produces.
location of their choice as illustrated in Fig. 4, the For future work, we would like to incorporate content and
implementation architecture. TCPD section of the figure contextual information for data trust prediction and propose
implements appropriate components of the framework as a more accurate prediction model based on artificial
shown in Fig. 2, for providing trusted data to the interested intelligence concepts. Although ITU-T has started a new
parties. This is a typical publish subscribe system whereby work on trust index which is a comprehensive accumulation
publishers publish the sensor data to the broker and of trust indicators to evaluate and quantify trust of entities,
subscribers receive notifications matching their until now, standards on trusted data are still very limited and
subscriptions from the broker. As illustrated in the Fig. 2, the current standards on entity or network based trust must be
TCPD can communicate with the IoT platform via an edge expanded for taking into consideration the data trust matters
server that implements the IGetTrustData and as explained in this work.
IProvideTrustData interfaces. Also, the TCPD can receive
data from the IoT platform for predicting the trust of such ACKNOWLEDGMENT
received data. This work was supported by Institute for Information &
Finally, Fig. 5 illustrates an example of a scaled down communications Technology Promotion(IITP) grant funded
sequences of interactions between some important by the Korea government(MSIT). [2015-0-00533,
stakeholders of an implementation instance of the system. Development of TII(Trusted Information Infrastructure)
Anytime a new environment sensor is available, it registers S/W Framework for Realizing Trustworthy IoT Eco-
its presence with the sensing broker, which in turn informs system].
the framework of the new available sensor. The new sensor
can then publish its data to the broker. The broker notifies REFERENCES
the TCPD to predict the trust of the received data. Similarly, [1] U. Jayasinghe, H. W. Lee, and G. M. Lee, “A Computational
whenever a new subscriber joins the system, its subscription Model to Evaluate Honesty in Social Internet of Things,” in
is submitted to the broker via the TCPD system. If a 32nd ACM SIGAPP Symposium On Applied Computing,
subscription matching at least one of the subscriptions of the Marrakesh, Morocco., 2017.

– 152 –