A highly adaptable probabilistic model for self-diagnosis of GPON-FTTH access network

Serge Romaric Tembo Mouafo (1, 2, 3) , Sandrine Vaton (1, 2) , Jean-Luc Courant (3) , Stephane Gosselin (3)

1 ADOPNET - Advanced technologies for operated networks
2 INFO - Département informatique
3 Orange Labs [Lannion]

Model-based approaches for self-diagnosing of telecommunication networks develop reasonings based on formal and explicit representation of network structure and network behavior. Network behavior modeling is a central issue for these methods. In a recent work, we have proposed a model of architecture and fault propagation of the FTTH (Fiber To The Home) access networks based on GPON (Gigabit capable Passive Optical Network). This model is based on a Bayesian network which encodes expert knowledge about the transport network and the connection network of subscribers. In this paper we extend this model by designing a model of the distribution network which fits to the various engineering techniques of the GPON-FTTH network. We carried out self-diagnosis of an operating GPON-FTTH network based on these two models. The performance of self-diagnostic of the new model is evaluated with respect to the previous model of the GPON-FTTH network.

Domaines: Intelligence artificielle [cs.AI]; Interface homme-machine [cs.HC]; Réseaux et télécommunications [cs.NI]

Serge Romaric Tembo Mouafo, Sandrine Vaton, Jean-Luc Courant, Stephane Gosselin. A highly adaptable probabilistic model for self-diagnosis of GPON-FTTH access network. SoftCOM 2016 : 24th International Conference on Software, Telecommunications and Computer Networks, Sep 2016, Split, Croatia. pp.1 - 5, ⟨10.1109/SOFTCOM.2016.7772106⟩. ⟨hal-01424652⟩