PdD Design of AI-Based Models for Fault Management in Optical Networks
Job title: PdD Design of AI-Based Models for Fault Management in Optical Networks
Company: Orange
Job description: about the roleYour role is to carry out a thesis work on: “The design of AI-based models for fault management in optical networks”. The increase in computing capacity and the huge amount of data available have greatly accelerated the use of artificial intelligence (AI) in different fields. In the context of optical communication systems, AI can provide innovative solutions related to improving network performance, network reliability and failure management. Fault management is an essential feature of the Network Management System (NMS) to avoid the degradation of the optical signal or even the interruption of the service. We distinguish two main families of failure management approaches.The first family allows reactive diagnosis. As soon as a failure occurs in the network, the operator must identify and localize its “root cause” as quickly as possible to allow the necessary interventions and thus reduce the impact of this failure on the deployed services. The objective of this first part of the thesis is to exploit data related to topology, performance and alarms by AI-based algorithms to perform a rapid diagnosis of the network and localize the fault. Classification techniques will be firstly used to create clusters containing correlated alarms (linked to the same source of failure). Secondly, AI-based algorithms will be developed to localize and identify the root cause of the failure. Finally, natural language processing techniques, generative AI, and Large Langage Models (LLM) will be studied to provide the results of the previous steps interactively with the field supervision operator and offer assistance in fault diagnosis.The second family of approaches seeks to predict the occurrence of failures due to a progressive degradation over time (e.g., aging of a device). These preventative approaches consist of detecting abnormal behavior by monitoring specific performance metrics. Through the analysis of the physical changes that the failures cause, the AI can learn the behavior of the optical system and predict failures. Thus, the NMS will be able to plan procedures for re-routing and re-optimization of the services likely to be affected. The objective of this second part of the thesis is to identify the metrics to be taken into account to predict the failure of equipment and then to define AI-based models to predict failures.The data are provided and comes from different sources: simulation data, experimental test-bed and field data.about youWe are looking for a Fresh Grad or early in career candidate that hold or be in the process of obtaining an Engineering, a Master or a University degree equivalent to a European Master’s (5-year duration) with a specialty in artificial intelligence, data science, telecommunication or a related discipline. Furthermore, the candidate will show the following competencies and skills:
- Strong knowledge on AI and machine learning,
- Good understanding of LLM(Large Langage Models),
- Familiarity with open source machine learning platforms like Tensorflow, PyTorch, Hugging Face,
- Proficient programming skills in Python,
- Interest in networking and optical communication,
- Open minded and synthetic with capability to step back from activities to propose new ideas,
- Autonomous, curious and proactive,
- Ability to communicate with experts and present results,
- Fluency in English is mandatory.
additional informationWithin the “Architecture and programmable Optical Transmission” (AOT) department, we are studying, within the framework of AION (Artificial Intelligence for Optical Networks) project, the potential of using AI to improve the functions of the optical transport network, in particular the management of failures and anomalies. In this context, we are conducting proof of concept demonstrations with several Orange affiliates with the aim of simplifying their preventive and curative maintenance operations. In order to train our algorithms, we use simulation data as well as real data collected in the live network. You will have the opportunity to work within a dynamic team of experts in optical networks and AI on a promising subject. Scientific publications and exchanges with Orange operational staff and international experts are planned during the thesis.departmentOrange Innovation brings together the research and innovation activities and expertise of the Group’s entities and countries. We work every day to ensure that Orange is recognized as an innovative operator by its customers and we create value for the Group and the Brand in each of our projects. With 740 researchers, thousands of marketers, developers, designers and data analysts, it is the expertise of our 6,000 employees that fuels this ambition every day.Orange Innovation anticipates technological breakthroughs and supports the Group’s countries and entities in making the best technological choices to meet the needs of our consumer and business customers. Within Orange Innovation, you will be integrated into a research team at the forefront of innovation and expertise in the fields of optical transmission, automation and control of transport networks. The team has several experimental platforms as well as test and simulation tools to recreate optical propagation conditions, generate data and evaluate the performance of the proposed solutions.contractThesis
Expected salary:
Location: Lannion, Côtes-d’Armor
Job date: Fri, 05 Jul 2024 22:04:36 GMT
Apply for the job now!
Submit your Resume!
