LEARNET: Reinforcement Learning Based Flow Scheduling for Asynchronous Deterministic Networks

Jonathan Prados-Garzon, Tarik Taleb, Miloud Bagaa

Tutkimustuotos: Artikkeli kirjassa/konferenssijulkaisussaConference contributionScientificvertaisarvioitu

18 Sitaatiot (Scopus)
282 Lataukset (Pure)


Time-Sensitive Networking (TSN) and Deterministic Networking (DetNet) standards come to satisfy the needs of many industries for deterministic network services. That is the ability to establish a multi-hop path over an IP network for a given flow with deterministic Quality of Service (QoS) guarantees in terms of latency, jitter, packet loss, and reliability. In this work, we propose a reinforcement learning-based solution, which is dubbed LEARNET, for the flow scheduling in deterministic asynchronous networks. The solution leverages predictive data analytics and reinforcement learning to maximize the network operator's revenue. We evaluate the performance of LEARNET through simulation in a fifth-generation (5G) asynchronous deterministic backhaul network where incoming flows have characteristics similar to the four critical 5GQoS Identifiers (5QIs) defined in Third Generation Partnership Project (3GPP) TS 23.501 V16.1.0. Also, we compared the performance of LEARNET with a baseline solution that respects the 5QIs priorities for allocating the incoming flows. The obtained results show that, for the scenario considered, LEARNET achieves a gain in the revenue of up to 45 compared to the baseline solution.

Otsikko2020 IEEE International Conference on Communications, ICC 2020 - Proceedings
ISBN (elektroninen)9781728150895
DOI - pysyväislinkit
TilaJulkaistu - kesäk. 2020
OKM-julkaisutyyppiA4 Artikkeli konferenssijulkaisussa
TapahtumaIEEE International Conference on Communications - Virtual Conference, Dublin, Irlanti
Kesto: 7 kesäk. 202011 kesäk. 2020


NimiIEEE International Conference on Communications
ISSN (painettu)1550-3607
ISSN (elektroninen)1938-1883


ConferenceIEEE International Conference on Communications


Sukella tutkimusaiheisiin 'LEARNET: Reinforcement Learning Based Flow Scheduling for Asynchronous Deterministic Networks'. Ne muodostavat yhdessä ainutlaatuisen sormenjäljen.

    Costa Requena, J.


    Projekti: Business Finland: Other research funding

  • 5G-FORCE-Taleb

    Taleb, T., Addad, R., Amor, A., Afolabi, I., Khennouche, H., Kerfah, I. & Batouche, A.


    Projekti: Business Finland: Other research funding

  • CSN: Customized Software Networking across Multiple Administrative Domains

    Taleb, T., Addad, R., Afolabi, I., Amor, A., Yu, H., Kianpisheh, S., Mariouak, M., Hellaoui, H., Sehad, N., Boudi, A., El Marai, O., Shokrnezhad, M., Bagaa, M., Maity, I., Naas, S., Bekkouche, O., Benzaid, C., Kerfah, I., Mada, B. & Yang, B.


    Projekti: Academy of Finland: Other research funding

Siteeraa tätä