Abstract

The Internet of Things (IoT) makes attractive services available to smart objects and humans. To aim this, IoT devices need high sensing, reasoning, and real-time acting capabilities that can be also obtained by promoting adaptive forms of cooperation machine-to-machine among smart objects. The convergence of IoT and multiagent systems also relies on the association between software agents with IoT devices for exploiting their social attitude of interacting and cooperating for services. However, the choice of reliable partners for cooperation can be very difficult when IoT devices migrate across different environments, where the most part of their members will be unreferenced with respect to their trustworthiness. It is well known that agents reputation can be a viable aspect to consider to form social groups; therefore, a possible solution to this problem is to form groups of agents in each IoT environment, based on their social capabilities. In this respect, the first contribution of this paper is represented by a reputation model focused on building the reputation capital of each agent. Second, an algorithm capable to form groups of agents in IoT environments on the basis of their reputation capital was designed. Finally, since in this contest, it is important to spread reliable and certified information about the device/agent reputation in a distributed environment, the third contribution is represented by the adoption of the blockchain technology to certify the reputation capital. Some experiments we have performed show that the model is capable to detect almost all the misleading agents if their percentage is under a high enough threshold, and that good results in term of group composition are obtained. Moreover, the simulations show that, by adopting our model, malicious devices always pay for services significantly more than honest ones. We argue that the individual reputation capital of devices and, consequently, the overall reputation capital of the IoT community can take benefit from the adoption of the proposed approach.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call