Doctoral thesis: « Machine Learning based localization in 5G »

Télécom SudParis

Doctoral School: Sciences et Technologies de l'Information et de la Communication and the Research Unit SAMOVAR (UMR 5157) - Services répartis, Architectures, Modélisation, Validation, Administration des Réseaux are presenting the "examination of a thesis" by Mr Abdallah SOBEHY who is expected to defend his research to obtain his PhD at l'Institut Polytechnique de Paris, prepared at Telecom SudParis in : Computer science

« Machine Learning based localization in 5G »

FRIDAY NOVEMBER 6, 2020 at 2:00 pm (Telecom SudParis - Room C06
9 rue Charles Fourier, 91000 Evry-Courcouronnes).

Jury members:

  • M. Eric RENAULT, Professor, ESIEE Paris, FRANCE - Supervisor
  • M. Paul MUHLETHALER, EVA Team Leader, Inria, FRANCE - Co-supervisor
  • M. Nadjib AIT SAADI, Full professor, Université de Versailles Saint-Quentin-en-Yvelines, FRANCE - Examinateur
  • M. Sidi-Mohammed SENOUCI, Professor, Université de Bourgogne, FRANCE - Rapporteur
  • Mme Hamida SEBA, Associate Professor, Université Claude Bernard Lyon 1, FRANCE - Rapporteur
  • M. Stephane MAAG, Professor, Telecom SudParis, FRANCE - Examiner
  • Mme Oyunchimeg SHAGDAR, Research Team Leader, VEDECOM, FRANCE - Examiner
  • M. Urko ZURUTUZA, Associate Professor, Mondragon University, SPAIN - Examiner


Localization is the process of determining the position of an entity in a local or global coordinate system. The applications of localization are widely spread across different contexts. For instance, in events, tracking the participants can save lives during crises. In health-care, elderly people can be tracked to respond to their needs in critical situations like falling. In warehouses, robots transferring products from one place to another require accurate knowledge of products' positions as well as other robots.

In industrial context of the factory of the future, localization is invaluable to achieve automated processes that are flexible enough to be reconfigured for various purposes. Localization is considered a topic of high interest both in the academia and industry especially with the advent of 5G. The requirements of 5G pave the way for revolutionizing localization capabilities; Enhanced Mobile Broadband (eMBB) that is expected to reach 10 Gbits/s, Ultra-Reliable Low-Latency Communication (URLLC) which is less than 1 ms and massive Machine-Type Communication (mMTC) allowing to connect around 1 million devices per km.

In this work, we focus on two main types of localization; range-based localization and fingerprinting based localization. In range-based localization, a network of devices with a maximum communication range estimate inter-distance values to their first-hop neighbors. These distances along with knowledge of positions of few anchor nodes are used to localize other nodes in the network using a triangulation based solution.

The proposed method is capable of localizing ≈ 90% of nodes in a network with an average degree of ≈ 10. In the second contribution, wireless channel responses, aka. Channel State Information (CSI) is used to estimate the position of a transmitter communicating with a MIMO antenna. In this work, we apply classical learning techniques (K-nearest neighbors) and deep learning techniques (Multi-Layer Perceptron Neural Network and Convolutional Neural Networks) to localize a transmitter in indoor and outdoor contexts.

Our work achieved the first place in the indoor positioning competition prepared by IEEE's Communication Theory Workshop among 8 teams from highly reputable universities worldwide by achieving a Mean Square Error of 2.3 cm.