A hybrid trust management framework for a multi-service social IoT network

Abstract

Social IoT (or SIoT) is an alternate architectural pattern for IoT, which involves IoT devices with social behavioural attributes. A SIoT-based service network makes use of social collaboration between IoT devices to enable low-latency collaborative services and applications. A key challenge in implementing a SIoT-based service network in a multi-vendor environment of heterogeneous devices is the issue of Trust. In this paper, we propose a hybrid trust management framework that makes use of Probabilistic Neighbourhood Overlap (P-NO), a method for estimating tie-strengths between nodes. The neighbourhood overlap concept is borrowed from past research in sociology and extended in our paper for directed social networks. Our proposed trust management framework is hybrid because: (1) P-NO is applied on a social graph that is generated from two types of social networks — the IoT device owners’ online social network (like Facebook) and the IoT-devices’ social network (i.e. the SIoT network). Accordingly, the approach makes use of both human intelligence and device artificial intelligence for trust management. (2) The framework uses a mix of dynamic (interaction-based) and static (graph-based) approach for trust management. It helps in limiting resource overheads of a pure dynamic approach, while still benefiting from its higher accuracy compared to a pure-static approach. We provide both theoretical and simulation-based analysis of our trust management framework. Our study shows the effectiveness of the proposed framework in handling different attack scenarios while requiring limited storage and computational resources in IoT devices.