BiLSTM-Based Fault Anticipation for Predictive Activation of FRER in Time-Sensitive Industrial Networks

Source abstract

Frame Replication and Elimination for Reliability (FRER) in Time-Sensitive Networking (TSN) enhances fault tolerance by duplicating critical traffic across disjoint paths. However, always-on FRER configurations impose constant redundancy overhead, even under nominal network conditions. This paper proposes a predictive FRER activation framework that leverages KPI-driven fault anticipation using a bidirectional Long Short-Term Memory (BiLSTM) model. By continuously analyzing multivariate performance indicators such as latency, jitter, and retransmission rates, the model forecasts impending faults and proactively triggers FRER. Redundancy is deactivated upon KPI recovery or after a minimum protection window, minimizing bandwidth consumption without compromising reliability. The framework includes a Python-based simulation environment, a real-time Streamlit dashboard for visualization, and a fully integrated runtime controller. Experimental results demonstrate significant improvements in link utilization while maintaining protection guarantees, showing the effectiveness of anticipatory redundancy strategies in industrial TSN environments.