TFL-DT: A Trust Evaluation Scheme for Federated Learning in Digital Twin for Mobile Networks

Due to the distributed collaboration and privacy protection features, federated learning is a promising technology perform model training in virtual twins of Digital Twin for Mobile Networks (DTMN). In order enhance reliability model, it always expected that users involved have trustworthy behaviors. Yet, available trust evaluation schemes problems considering simplex factor using coarse-grained calculation method. this paper, we propose scheme DTMN, which takes direct evidence recommended information into account. A user behavior designed based on multiple attributes depict users’ fine-grained manner. Furthermore, methods local value are proposed data as evidence. Several experiments were conducted verify effectiveness scheme. The results show method able evaluate levels with different patterns accurately. Moreover, performs better resisting attacks from alternately execute good bad behaviors compared state-of-the-art