GBTrust: Leveraging Edge Attention in Graph Neural Networks for Trust Management in P2P Networks

Trust is an important factor in the success of P2P networks. It is needed to ensure that nodes in the network can be trusted to behave honestly and to deliver on their promises (e.g sharing resources). While traditional reputation trust management systems (RTMS) such as BTrust or EigenTrust have proven effective, there is room for further enhancement by integrating advanced graph neural network (GNN) models. This paper proposes a novel approach to enhance Trust Management Systems (TMS) by incorporating an Edge-Feature Attention Mechanism into the Edge Graph Neural Network (EGNN) model, which takes into account the direction of edges. The proposed GBTrust model is specifically designed for trust management in P2P networks, leveraging the interactions and relationships among peers. By incorporating the Edge-Feature Attention Mechanism, the model dynamically assigns importance to different edge features based on their relevance, thereby improving the discrimination of the importance of various neighbors and edge features in the network graph. The GBtrust model aims to provide more accurate and adaptive detection of malicious peers, thereby enhancing the overall security and reliability of P2P networks.