PhD "Energy savings with Machine Learning for 6G MIMO" F/M
Orange
Châtillon, Hauts-de-Seine
CDI
Temps-plein
Il y a 2 mois
about the roleYour role is to carry out a PhD thesis work on “Energy savings with Machine Learning for 6G MIMO”.We are witnessing a craze, particularly among telecom equipment manufacturers, in the use of Machine Learning (ML, AI techniques) to design mobile communications systems. In order to be able to compare the solutions best suited to its needs, Orange must understand how ML techniques work. Developing solutions based on ML will make it possible to operate networks efficiently, and to “challenge” equipment manufacturers on their ML-based products.In this context and unlike classic Deep Learning models such as Recurrent Neural Networks, so-called “liquid” neural network models work in continuous time and benefit from explicit analytical expressions. They were recently developed by a team from MIT and successfully applied to problems of estimating probability densities and modeling data with complex dynamics (irregularly sampled data), for use cases of automatic drone navigation for example.We will apply these innovative models to MIMO and Beamforming, essential building blocks for 6G, and will contribute to the specification for future radio networks combining the traditional performance/complexity compromise with energy efficiency criteria, giving a prominent place to those, in order to respond to the societal and environmental challenges of the Orange Group.This thesis is part of Orange's strategic focus on Networks of the Future: remaining at the forefront of network developments and their operations to develop our core business.The aim of the thesis is to implement MIMO and Beamforming systems for 6G networks via new AI techniques, making it possible to better modulate the trade-off between performance/complexity and energy consumption.We will address the MIMO detection problem, requiring the development of suboptimal solutions to be deployed in operational networks. In practice, approaches based on “liquid” neural networks will therefore be able to challenge classic Deep Learning models (DetNet) and be compared with traditional detection algorithms. For Beamforming issues, we will address the optimization of the precoding matrix, in a realistic framework under power, EMF or minimum flow constraints.The challenge is to combine the traditional performance/complexity compromise (BER/SNR, and theoretical FLOPS or measured CPU calculation times) of resolution techniques with the energy efficiency requirements targeted for future 6G networks. Thus, a critical analysis of current communications systems from the point of view of energy consumption will be carried out as part of the state of the art to be performed during the thesis.about you- Knowledge and experience in the field of mobile digital communications (signal processing).- Knowledge and experience in the field of Machine Learning and Data Science.- Knowledge and experience in Deep Learning would be a plus.- Curiosity, rigor, autonomy.- Master and/or Engineering Degree.additional informationThe innovative approach here consists of building systems based not on standard Deep Learning (DL), but on so-called “liquid” artificial neural networks. Indeed, these recent models are compact and easier to train than those from classic DL: reduced number of parameters and memory consumption.You will join a dynamic and multicultural research and R&D team with internationally recognized radio and optimization/learning experts. You will evolve in an innovation team whose prerogatives are at the heart of the deployment of new generations of mobile communications (5G/6G).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 720 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 the “Innovation in Radio and Environment” department of Orange Innovation, you will be integrated into a research team at the forefront of innovation and expertise on the networks of the future (5G/6G). You will be part of a research ecosystem alongside experts in mathematical modeling and data science whose main missions are to develop advanced techniques based on Artificial Intelligence, to model and optimize the costs of network infrastructures, and to design mathematical models and tools to support strategic decisions. You will be required to interact with design engineers in anticipation to enable the concrete implementation of the concepts studied, benefiting from 5G/6G simulation platforms.contractThesis
