Summary
Network problems can cause unexpected and hard-to-predict drops in performance. The paper’s authors noticed that existing methods either ignore how these problems spread or give only fixed point guesses without showing uncertainty. They created a new way to model fault spreading using concepts from information theory like entropy and mutual information. They also built a special system called FEMNet that uses this approach to better predict how network performance might change when faults happen, including estimating uncertainties. Overall, their method improves the accuracy of both precise and probabilistic predictions about network issues.
network faultfault propagationinformation entropymutual informationtransfer entropygraph message-passingconditional diffusion modelperformance predictionprobabilistic modelingKPI
Authors
Lusha Mo, Fengxiao Tang, Xiaonan Wang, Ming Zhao
Abstract
Network faults can trigger cascading effects that cause abrupt and nonstationary performance degradation. Existing learning-based performance predictors mainly focus on normal operation or treat fault-induced topology and routing changes as static inputs, and typically produce deterministic point estimates. They overlook fault-propagation dynamics and uncertainty in performance evolution. The predefined-rule and purely data-driven propagation models lack a unified representation of fault definition, propagation mechanism, and impact quantification. Additionally, generic denoisers in conditional diffusion models fail to incorporate fault propagation into uncertainty modeling. To address these limitations, we propose an information-entropy-driven fault propagation paradigm (IEFP) that characterizes fault propagation via relative entropy, mutual information and transfer entropy. We then design a fault-aware graph message-passing mechanism that propagation contexts modulate network representation learning. We further develop FEMNet, which employs this mechanism as a tailored denoiser within a conditional diffusion model to enable probabilistic network performance prediction under complex fault scenarios. Compared with the strongest baselines, IEFP improves fault-prediction performance, while FEMNet reduces errors in both point and probabilistic KPI prediction.