Trust-aware game-theoretic allocation for secure and efficient IoT-edge systems

Official URL: https://dx.doi.org/10.1109/ICEDGE67252.2025.11412888

The rapid expansion of Internet of Things (IoT) ecosystems demands secure and efficient service allocation in edge computing environments. While game-theoretic models offer strong potential for resource optimization, most fail to address the risks posed by malicious or non-cooperative devices. In this study, we introduce a trust-aware, many-to-one matching game that integrates a dynamic trust mechanism into the allocation process. Trust scores, updated based on observed device behavior, influence both entity and server utility functions, enabling the autonomous detection and rejection of untrustworthy tasks. The proposed algorithm is evaluated across two distinct scenarios: a real-world Intel Berkeley sensor network and a synthetic Azure IoT Edge architecture. Experimental results show that our model achieves 100% malicious task rejection, superior load balancing, and high Quality of Service for legitimate users, significantly outperforming Greedy and Random baselines. These findings demonstrate that embedding dynamic trust into game-theoretic allocation models offers a robust strategy for building secure and resilient edge computing systems. The source code and supplementary materials supporting this study are publicly accessible via our GitHub repository.