This paper presents a self-supervised machine learning approach for cognitive reconfiguration in a Hyper-X-like flexible-bandwidth optical interconnect architecture. The proposed approach makes use of a clustering algorithm to learn the traffic patterns from historical traces. A heuristic algorithm is developed for optimizing the connectivity graph for each identified traffic pattern. Further, to mitigate the scalability issue induced by frequent clustering operations, we parameterize the learned traffic patterns by a deep neural network classifier. The classifier is trained offline by supervised learning to enable classification of traffic matrices during online operations, thereby facilitating cognitive reconfiguration decision making. Simulation results show that compared with a static all-to-all interconnection, the proposed approach can improve throughput by up to 1.76× while reducing end-to-end packet latency and flow completion time by up to 2.8× and 25×, respectively.

Reconfigurable Optical Datacom Networks by Self-supervised Learning / Liu, C. -Y.; Chen, X.; Proietti, R.; Li, Z.; Yoo, S. J. B.. - ELETTRONICO. - (2021), pp. 23-27. (Intervento presentato al convegno 2021 ACM SIGCOMM 2021 Workshop on Optical Systems, OptSys 2021 tenutosi a USA nel 23 August 2021) [10.1145/3473938.3474509].

Reconfigurable Optical Datacom Networks by Self-supervised Learning

Proietti R.;
2021

Abstract

This paper presents a self-supervised machine learning approach for cognitive reconfiguration in a Hyper-X-like flexible-bandwidth optical interconnect architecture. The proposed approach makes use of a clustering algorithm to learn the traffic patterns from historical traces. A heuristic algorithm is developed for optimizing the connectivity graph for each identified traffic pattern. Further, to mitigate the scalability issue induced by frequent clustering operations, we parameterize the learned traffic patterns by a deep neural network classifier. The classifier is trained offline by supervised learning to enable classification of traffic matrices during online operations, thereby facilitating cognitive reconfiguration decision making. Simulation results show that compared with a static all-to-all interconnection, the proposed approach can improve throughput by up to 1.76× while reducing end-to-end packet latency and flow completion time by up to 2.8× and 25×, respectively.
2021
9781450386500
File in questo prodotto:
File Dimensione Formato  
3473938.3474509.pdf

non disponibili

Tipologia: 2a Post-print versione editoriale / Version of Record
Licenza: Non Pubblico - Accesso privato/ristretto
Dimensione 930.35 kB
Formato Adobe PDF
930.35 kB Adobe PDF   Visualizza/Apri   Richiedi una copia
OptSys_2021 (1).pdf

accesso aperto

Tipologia: 2. Post-print / Author's Accepted Manuscript
Licenza: PUBBLICO - Tutti i diritti riservati
Dimensione 983.88 kB
Formato Adobe PDF
983.88 kB Adobe PDF Visualizza/Apri
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11583/2973498