A Reputation based Trust Management System for the Internet of Things
Conference : International Symposium on Networked Cyber-Physical Systems (NetCPS)
Date: September 19 - September 20, 2016
The Internet-of-Things is gradually being integrated within Future Internet service delivery models and utility based cloud-computing paradigms. To date a number of open service frameworks (e.g., OpenIoT, Cosm etc.) have been implemented for building/deploying and managing IoT applications.
In this particular study we propose a reputation based trust framework as a means to assess the trustworthiness of interconnected heterogeneous objects (e.g. sensors, smart objects etc.), either at the operational level (e.g. is the object performing its intended operation), or at the data level, (e.g. is the data timely and accurate delivered). The goal is: (i) to assist IoT application service providers in the selection of such objects for building trustworthy IoT applications; (ii) empower consumers to select objects and/or data emanating from them, based on dynamic performance metrics.
Reputation-based trust management systems represent a considerable trend in decision support for service provision applications. Reputation mechanisms are also employed in Web search engines (i.e. Google’s PageRank, TrustRank), online marketplaces – e-Commerce (i.e. eBay, Amazon), email systems to provide anti-spam functionality, etc. Straightforward reputation schemas mostly based on users’ voting, can also be found in experts’ communities such as Slashdot and StackExchange.
The mechanisms for building up the reputation of interconnected objects are based on the combination of Quality of Experience (QoE), perceived as end user’s feedback upon using services that utilize data streams from these contributing sources, and monitoring data, in the case that such an information is available. End-user’s feedback regarding the service is delegated to the actual smart objects / data streams that the service comprises of. Reputation values are calculated: (a) per sensing resource, in the case of deployed sensor networks; and (b) per contributor, in the case of volunteering resources (e.g., smartphone).Download Full Paper