pr\u00e9sentent<\/p>\n\n\n\n
Autoris\u00e9 \u00e0 pr\u00e9senter ses travaux en vue de l\u2019obtention du Doctorat de l’Institut Polytechnique de Paris, pr\u00e9par\u00e9 \u00e0 T\u00e9l\u00e9com SudParis en :<\/p>\n\n\n\n
le MARDI 12 D\u00c9CEMBRE 2023 \u00e0 9h00<\/p>\n\n\n\n
\u00e0<\/p>\n\n\n\n
B206
9 Rue Charles Fourier, 91000 EVRY, France<\/p>\n\n\n\n
Membres du jury :<\/strong><\/p>\n\n\n\n M. Noel CRESPI<\/strong>, Full professor, T\u00e9l\u00e9com SudParis, FRANCE – Directeur de these R\u00e9sum\u00e9 :<\/strong><\/p>\n\n\n\n L’estimation de la qualit\u00e9 de transmission (QoT) des chemins optiques est cruciale dans la conception du r\u00e9seau et le provisionnement des services. Des \u00e9tudes r\u00e9centes se sont tourn\u00e9es vers les techniques de l’intelligence artificielle (IA) pour am\u00e9liorer la pr\u00e9cision de l’estimation de la QoT, en utilisant les donn\u00e9es g\u00e9n\u00e9r\u00e9es par le r\u00e9seau optique. Nous distinguons quatre cat\u00e9gories de solutions. La premi\u00e8re cat\u00e9gorie consiste \u00e0 construire un mod\u00e8le d\u2019IA pour v\u00e9rifier la faisabilit\u00e9 d’un chemin optique. La deuxi\u00e8me cat\u00e9gorie vise \u00e0 proposer des mod\u00e8les bas\u00e9s sur l’IA pour remplacer les mod\u00e8les analytiques. La troisi\u00e8me cat\u00e9gorie utilise l’IA pour am\u00e9liorer les performances des mod\u00e8les d’estimation de la QoT en r\u00e9duisant l’incertitude sur les param\u00e8tres d’entr\u00e9e. La derni\u00e8re cat\u00e9gorie consiste \u00e0 am\u00e9liorer les performances et la capacit\u00e9 de g\u00e9n\u00e9ralisation des solutions \u00e0 base d’IA en am\u00e9liorant les \u00e9chantillons des jeux de donn\u00e9es dans la phase d’apprentissage gr\u00e2ce \u00e0 des techniques d’apprentissage par transfert. Les mod\u00e8les d’estimation de la QoT peuvent constituer un module dans le Digital Twin du r\u00e9seau optique, visant \u00e0 simuler l’impact d’une nouvelle configuration sur la performance du r\u00e9seau avant la phase de d\u00e9ploiement. Cependant, ces mod\u00e8les requirent une connaissance parfaite de l’\u00e9tat du r\u00e9seau, repr\u00e9sent\u00e9 \u00e0 partir d\u2019un ensemble de param\u00e8tres optiques ayant des valeurs qui peuvent \u00eatre certaines ou incertaines. Les mesures de performance collect\u00e9es par le contr\u00f4leur peuvent repr\u00e9senter un feedback sur la pr\u00e9cision de l’estimation de la QoT, ce qui peut d\u00e9clencher des algorithmes \u00e0 base de machine learning pour raffiner les valeurs des param\u00e8tres incertains. Dans cette th\u00e8se, nous \u00e9tudions le probl\u00e8me d’incertitude des param\u00e8tres, et nous proposons trois approches pour am\u00e9liorer la QoT dans ce cas. Nous proposons pour chaque approche un certain nombre de processus d’apprentissage et nous testons leur performances avec des donn\u00e9es de simulation et des donn\u00e9es collect\u00e9es \u00e0 partir du r\u00e9seau op\u00e9rationnel. La premi\u00e8re approche se base sur l’optimisation des param\u00e8tres du r\u00e9seau en utilisant l’erreur dans l’estimation de la QoT comme fonction d’objectif. Nous impl\u00e9mentons cette approche avec deux processus d’apprentissage, le premier bas\u00e9 sur un mod\u00e8le analytique (GNPy) et le deuxi\u00e8me sur un mod\u00e8le \u00e0 base de machine learning (r\u00e9seau de neurones). Cette approche arrive \u00e0 minimiser l’erreur d’estimation jusqu’\u00e0 0~dB pour des configurations de r\u00e9seau o\u00f9 le mod\u00e8le a \u00e9t\u00e9 d\u00e9j\u00e0 entra\u00een\u00e9, et atteint une erreur d’estimation de 0.3~dB sur des nouvelles configurations. Dans la deuxi\u00e8me approche, nous r\u00e9-entra\u00eenons un mod\u00e8le bas\u00e9 sur un r\u00e9seau de neurones pour l’adapter \u00e0 l’incertitude des param\u00e8tres en utilisant l’apprentissage par transfert. Nous montrons que le mod\u00e8le peut apprendre un nouveau comportement sans optimiser les param\u00e8tres incertains. Nous arrivons \u00e0 une erreur de validation de 0.5~dB avec seulement dix nouveaux \u00e9chantillons. La derni\u00e8re approche consiste \u00e0 d\u00e9tecter les changements de param\u00e8tres en r\u00e9ponse \u00e0 un \u00e9v\u00e9nement de panne en utilisant des techniques d\u2019apprentissage par renforcement. Nous consid\u00e8rons deux types de pannes, et nous montrons que le mod\u00e8le atteint une pr\u00e9cision de classification de 93% dans une petite topologie. Enfin, nous appliquons la premi\u00e8re approche bas\u00e9e sur l’optimisation Bay\u00e9sienne pour raffiner les param\u00e8tres du r\u00e9seau sur des donn\u00e9es collect\u00e9es \u00e0 partir d\u2019un r\u00e9seau op\u00e9rationnel. Nous extrayons les donn\u00e9es \u00e0 partir de l’interface nord du contr\u00f4leur, et nous construisons l’\u00e9tat du r\u00e9seau selon le mod\u00e8le de donn\u00e9e d\u2019entr\u00e9e de GNPy. En appliquant notre processus d’apprentissage sur deux lignes de transmission, nous constatons une r\u00e9duction d\u2019erreur qui atteint 1.7~dB sur les services monitor\u00e9s.<\/p>\n\n\n\n Abstract :<\/strong><\/p>\n\n\n\n Estimating lightpath Quality of Transmission (QoT) is crucial in network design and service provisioning. Recent studies have turned to artificial intelligence (AI) techniques to improve the accuracy of QoT estimation using the data generated by the operational network. We distinguish four categories of solutions. The first category consists of building AI models to check the feasibility of a lightpath. The second category aims to predict the exact QoT performance in order to compete with analytical models. The third category uses AI to improve the performance of QoT estimation models by reducing the uncertainty on input parameters. The last category consists of improving the performance and generalization ability of AI-based solutions by retraining the models using the least amount of training samples through transfer learning techniques. QoT models can act as part of the digital twin of the operational network by simulating the impact of new network configurations before deploying them. However, they require a perfect knowledge of the network state, consisting of a set of optical parameters that have different levels of uncertainty. Using the QoT measurements collected by the network controller, we can have a feedback about the QoT estimation inaccuracy, which can potentially be addressed using ML based techniques. In this thesis, we study this issue of uncertainty in network parameters and consider three approaches that can improve the QoT estimation in this case. We propose different learning processes in each approach, and test their performance using simulation and real data. The first approach relies on optimizing the network parameters using the QoT estimation error as an objective function. We apply this approach through two learning processes to target QoT estimation tools based respectively on analytical model (GNPy) and Machine Learning (neural network). This approach can minimize the SNR estimation error to close to 0~dB on already trained network configuration, and reaches 0.3~dB estimation error on unseen network configurations. In the second approach, we retrain a neural network based model to adapt it to changes in QoT due to parameters uncertainty through Transfer Learning. We show how the model can relearn the new behavior of the network without searching for the correct values of the network parameters. We can reach up to 0.5~dB in validation error with only ten new training samples. The last approach consists of detecting parameter changes in response to failure events using reinforcement learning techniques. We consider two types of failure events. We show that the model can correctly classify the events with up to 93% of accuracy in small network topologies. Finally, we apply the first approach based on Bayesian Optimization algorithm to refine network parameters using data collected from a live network. We use data extracted from the north-bound interface of the network controller to build a network state based on the input data model of GNPy. Then, we apply our learning process on two transmission lines, which led to SNR estimation improvement up to 1.7~dB for the monitored services.<\/p>\n","protected":false},"excerpt":{"rendered":" L’Ecole doctorale : Ecole Doctorale de l’Institut Polytechnique de Paris et le Laboratoire de recherche SAMOVAR – Services r\u00e9partis, Architectures, Mod\u00e9lisation, Validation, Administration des R\u00e9seaux pr\u00e9sentent l\u2019AVIS DE SOUTENANCE de Monsieur Reda AYASSI Autoris\u00e9 \u00e0 pr\u00e9senter ses travaux en vue de l\u2019obtention du Doctorat de l’Institut Polytechnique de Paris, pr\u00e9par\u00e9 \u00e0 T\u00e9l\u00e9com SudParis en : […]<\/p>\n","protected":false},"author":4,"featured_media":0,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"ocean_post_layout":"","ocean_both_sidebars_style":"","ocean_both_sidebars_content_width":0,"ocean_both_sidebars_sidebars_width":0,"ocean_sidebar":"0","ocean_second_sidebar":"0","ocean_disable_margins":"enable","ocean_add_body_class":"","ocean_shortcode_before_top_bar":"","ocean_shortcode_after_top_bar":"","ocean_shortcode_before_header":"","ocean_shortcode_after_header":"","ocean_has_shortcode":"","ocean_shortcode_after_title":"","ocean_shortcode_before_footer_widgets":"","ocean_shortcode_after_footer_widgets":"","ocean_shortcode_before_footer_bottom":"","ocean_shortcode_after_footer_bottom":"","ocean_display_top_bar":"default","ocean_display_header":"default","ocean_header_style":"","ocean_center_header_left_menu":"0","ocean_custom_header_template":"0","ocean_custom_logo":0,"ocean_custom_retina_logo":0,"ocean_custom_logo_max_width":0,"ocean_custom_logo_tablet_max_width":0,"ocean_custom_logo_mobile_max_width":0,"ocean_custom_logo_max_height":0,"ocean_custom_logo_tablet_max_height":0,"ocean_custom_logo_mobile_max_height":0,"ocean_header_custom_menu":"0","ocean_menu_typo_font_family":"0","ocean_menu_typo_font_subset":"","ocean_menu_typo_font_size":0,"ocean_menu_typo_font_size_tablet":0,"ocean_menu_typo_font_size_mobile":0,"ocean_menu_typo_font_size_unit":"px","ocean_menu_typo_font_weight":"","ocean_menu_typo_font_weight_tablet":"","ocean_menu_typo_font_weight_mobile":"","ocean_menu_typo_transform":"","ocean_menu_typo_transform_tablet":"","ocean_menu_typo_transform_mobile":"","ocean_menu_typo_line_height":0,"ocean_menu_typo_line_height_tablet":0,"ocean_menu_typo_line_height_mobile":0,"ocean_menu_typo_line_height_unit":"","ocean_menu_typo_spacing":0,"ocean_menu_typo_spacing_tablet":0,"ocean_menu_typo_spacing_mobile":0,"ocean_menu_typo_spacing_unit":"","ocean_menu_link_color":"","ocean_menu_link_color_hover":"","ocean_menu_link_color_active":"","ocean_menu_link_background":"","ocean_menu_link_hover_background":"","ocean_menu_link_active_background":"","ocean_menu_social_links_bg":"","ocean_menu_social_hover_links_bg":"","ocean_menu_social_links_color":"","ocean_menu_social_hover_links_color":"","ocean_disable_title":"default","ocean_disable_heading":"default","ocean_post_title":"","ocean_post_subheading":"","ocean_post_title_style":"","ocean_post_title_background_color":"","ocean_post_title_background":0,"ocean_post_title_bg_image_position":"","ocean_post_title_bg_image_attachment":"","ocean_post_title_bg_image_repeat":"","ocean_post_title_bg_image_size":"","ocean_post_title_height":0,"ocean_post_title_bg_overlay":0.5,"ocean_post_title_bg_overlay_color":"","ocean_disable_breadcrumbs":"default","ocean_breadcrumbs_color":"","ocean_breadcrumbs_separator_color":"","ocean_breadcrumbs_links_color":"","ocean_breadcrumbs_links_hover_color":"","ocean_display_footer_widgets":"default","ocean_display_footer_bottom":"default","ocean_custom_footer_template":"0","ocean_post_oembed":"","ocean_post_self_hosted_media":"","ocean_post_video_embed":"","ocean_link_format":"","ocean_link_format_target":"self","ocean_quote_format":"","ocean_quote_format_link":"post","ocean_gallery_link_images":"off","ocean_gallery_id":[],"footnotes":""},"categories":[286,402],"tags":[],"class_list":["post-6404","post","type-post","status-publish","format-standard","hentry","category-fractualites-ennews-fr","category-seminaires-ness-2013-fr","entry"],"_links":{"self":[{"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/posts\/6404","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/users\/4"}],"replies":[{"embeddable":true,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/comments?post=6404"}],"version-history":[{"count":1,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/posts\/6404\/revisions"}],"predecessor-version":[{"id":6405,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/posts\/6404\/revisions\/6405"}],"wp:attachment":[{"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/media?parent=6404"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/categories?post=6404"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/samovar.telecom-sudparis.eu\/index.php\/wp-json\/wp\/v2\/tags?post=6404"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
M. Massimo TORNATORE<\/strong>, Full professor, Politecnico di Milano, ITALIE – Rapporteur
M. John PUENTES<\/strong>, Full professor, IMT Atlantique, FRANCE – Rapporteur
M. Yvan POINTURIER<\/strong>, Ing\u00e9nieur de recherche, Huawei Technologies France, FRANCE – Examinateur
M. Michel MORVAN<\/strong>, Ma\u00eetre de conf\u00e9rences, IMT Atlantique, FRANCE – Examinateur
Mme Catherine LEPERS<\/strong>, Full professor, Telecom SudParis, FRANCE – Examinateur
M. Ahmed TRIKI<\/strong>, Ing\u00e9nieur de recherche, Orange Innovation, FRANCE – Co-encadrant de these<\/p>\n\n\n\n\u00ab Techniques de l\u2019intelligence artificielle pour l\u2019am\u00e9lioration des performances et l\u2019optimisation des ressources des r\u00e9seaux optiques \u00bb<\/h2>\n\n\n\n
pr\u00e9sent\u00e9 par Monsieur Reda AYASSI<\/h2>\n\n\n\n