Ezmiron Deniku
Robust Early Detection of Cable Faults in National Telecommunication Networks.
Rel. Paolo Garza, Simone Monaco, Pino Castrogiovanni. Politecnico di Torino, Corso di laurea magistrale in Ingegneria Informatica (Computer Engineering), 2026
Abstract
Telecommunication networks generate very large alarm streams from heterogeneous devices distributed across wide geographic areas. Because a single physical incident may trigger many correlated events, operators first group alarms into clusters and then investigate their root cause. This thesis addresses the early identification of cable faults, a rare but high-impact class of physical failures that requires on-site intervention and causes substantial service degradation. The work was developed in collaboration with FiberCop S.p.A. and extends a previous autoencoder-based approach that represented each cluster as an unordered aggregate. In the following work, every cluster is modeled as an ordered sequence of alarms observed during the first 300 seconds after the cluster begins.
Categorical attributes are embedded, numerical attributes are projected into the model space, and timing is represented through relative-time features and Time2Vec encodings
Relatori
Anno Accademico
Tipo di pubblicazione
Numero di pagine
Informazioni aggiuntive
Corso di laurea
Classe di laurea
URI
![]() |
Modifica (riservato agli operatori) |
