ADAPTIVE MONITORING OF COGNITIVE OPTICAL NETWORKS TO ADDRESS THE PROBLEM OF LIMITED INFRASTRUCTURE INTEGRITY VISIBILITY

Abstract

The paper examines the relationship between optical network maintenance and security management systems. Changes in monitoring efficiency requirements and the uninterrupted operation of infrastructure are analyzed. Based on the study of anomalies associated with intentional and unintentional damage in PON networks, a cognitive monitoring algorithm is proposed that integrates spectral analysis of optical signals with machine learning methods. The proposed approach enhances the accuracy of fiber-optic infrastructure diagnostics, reduces the impact of weakly attenuated modes, and enables real-time fiber degradation prediction. Simulation results demonstrate that the use of multispectral analysis and adaptive anomaly classification methods can reduce fault detection time by 40–50%, lower maintenance costs by 30%, and decrease the frequency of critical failures by 35–40%. The proposed solution contributes to the development of more reliable and secure PON networks capable of adaptively responding to both technical and cyber threats.