Chair of Communications

Predictive Maintenance


Fiber-optic metro and core networks build the spine of today’s digital economy and society. Due to their unbeatable capacity these optical networks play a key role for the high-bitrate connectivity both in public and private areas. High requirements on reliability, security and sustainability call for a very high efficiency on the one hand, on the other hand also a high level on resilience is desired.

Failures in such complex networks are often difficult to localize and a long time is typically needed for fixing and replacement of the malfunctioning components. A proactive failure management, which can identify potential problems even before a major system disruption occurs, is not implemented in most cases today.

That is why we investigate how a lifetime prediction of the individual network components can be made with high accuracy using artificial neural networks (ANN). The results show that ANN-based predictions outperform conventional lifetime projections using accelerated aging tests significantly.


in cooperation with         ADVA