Machine to machine trust in the IoT era

Abstract

Machine to machine communications are at the center stage of the Internet of things (IoT). Connecting the physical world with the digital world not only creates new opportunities for innovation and discovery, but also opens doors for misuse and abuse. This paper argues that reputation based trust can be an effective countermeasure for securing machine-to-machine communications. We propose to establish machine-to-machine trust by taking into account both transaction/interaction service behaviors and feedback rating behaviors in the presence of bogus transactions and dishonest feedbacks. Our machine-to-machine trust model, called M2MTrust, introduces two novel trust metrics: (1) pairwise similarity based feedback credibility and (2) threshold-controlled trust propagation. We compute the direct trust from machine A to machine B by utilizing their pairwise rating similarity as the weight to the normalized aggregate of ratings that A has given to B. Our direct trust computation model can effectively constrain malicious nodes to gain direct trusts from dishonest feedback ratings by leveraging feedback credibility. Furthermore, our threshold-controlled trust propagation mechanism can successfully block the trust propagation from good nodes to malicious nodes. We conduct extensive experiments using simulation and real datasets and the experimental results show that M2MTrust significantly outperforms other trust metrics in terms of both attack resilience and performance in the presence of dishonest feedbacks and sparse feedback ratings against four representative attack models.