An Investigation of Trust Management Systems in Digital Twin Architectures

As the Internet of Things (IoT) continues to expand, Digital Twin (DT) architectures are adopted to represent physical devices with virtual entities, enabling scalable and decentralized systems. In this context, Trust Management Systems (TMSs) have been introduced to assess the trustworthiness of entities, supporting decisions in service exchange interactions. However, most existing TMSs assume ideal conditions, overlooking issues such as service evaluation inaccuracies and unintentional provider faults. This paper investigates the limitations of current TMSs in DT architectures by analyzing their structure and evaluating their performance under scenarios with errors. A classification of state-of-the-art trust models is first presented, followed by a general mathematical formulation that depicts the main inputs of trust computation. Then, several representative TMSs are simulated and evaluated under scenarios characterized by malicious behaviors and errors, both in service provisioning and evaluation. The results highlight important deficiencies in current designs, particularly in distinguishing between faulty but benevolent nodes and malicious entities.