A Scalable Subjective Logic-Based Reinforcement Learning Approach for Two-Way Intelligent Trust Management in Fog/Edge Computing

Mobile Edge Computing (MEC) and fog/edge computing provide computing and storage resources close to end users, making them suitable for latency-sensitive applications. In these environments, Service Function Chaining (SFC) enables network services by connecting multiple Virtual Network Functions (VNFs). However, most existing SFC placement methods mainly optimize latency, resource utilization, and deployment cost, while paying limited attention to the trustworthiness and reliability of edge nodes. As a result, services may be deployed on unreliable or compromised nodes, reducing service quality and continuity. To address this issue, this paper proposes SRTPP-SFC, a scalable trust- and reliability-aware framework for parallel SFC placement with intelligent VNF sharing. The framework introduces a bidirectional trust mechanism that allows both service requesters and service providers to evaluate each other’s trustworthiness before service establishment. Trust values are computed using a Subjective Logic model that combines direct interactions with indirect recommendations. These trust scores, together with node reliability, resource availability, and network conditions, are incorporated into an Actor–Critic reinforcement learning framework to guide adaptive placement decisions. The framework also supports parallel SFC deployment, reliability-aware backup provisioning, and LSTM-based VNF sharing and reuse to improve resource efficiency and service performance. Experimental results demonstrate that SRTPP-SFC achieves higher service acceptance, reliability, and trustworthiness than existing methods while maintaining low end-to-end latency and efficient resource utilization. These results indicate that the proposed framework provides an effective and scalable solution for secure and reliable SFC deployment in dynamic fog/edge computing environments.