16 May 2023
- Orange Innovation
- Research Field
- Computer science » Informatics
- Researcher Profile
- Recognised Researcher (R2)
Leading Researcher (R4)
First Stage Researcher (R1)
Established Researcher (R3)
- Application Deadline
- 14 Jun 2023 – 22:00 (UTC)
- Type of Contract
- Job Status
- Is the job funded through the EU Research Framework Programme?
- Not funded by an EU programme
- Is the Job related to staff position within a Research Infrastructure?
Your role is to carry out a PhD thesis on “AI for reliable and available wireless mesh networks for the Industry of the Future”. The objective is to study and propose Machine Learning algorithms, more specifically reinforcement learning, to perform resource allocation and scheduling of radio transmissions in an IEEE 802.15.4 mesh network implementing a 6TiSCH protocol stack (IETF standard protocols).
These networks implement the waveforms and radio modulations of the IEEE 802.15.4g standard, the channel access methods of the IEEE 802.15.4e TSCH standard, and an IETF-standardized IPv6 stack offering basic resource reservation mechanisms, IPv6 addressing and routing (RPL). They provide the necessary connectivity for devices used in Industrial IoT (IIoT) networks, in particular sensors and actuators that need to be operated on a stack, and their integration in non-critical processes of production chains is already enabling a deep transformation of the industry. However, the adoption of these networks for use cases requiring a high quality of service (Industry 4.0) remains limited, due to the radio interference existing in unlicensed frequency bands, and to the scheduling mechanisms used which do not allow to guarantee a delay or a data delivery rate.
In this context, the main objective of this thesis is to define a comprehensive toolkit to orchestrate communications in a wireless mesh network through a combination of centralized and distributed decisions based on reinforcement learning algorithms to meet application requirements.
To achieve this goal, you will study allocation and scheduling algorithms based on reinforcement learning and their application to wireless mesh networks. Specifically, Deep Q Learning (DQN) algorithms for centralized long-term resource allocation, and Multi-Armed Bandit (MAB) algorithms for connectivity restoration in case of topology changes as well as for continuous network optimization.
The main challenge is to model the network constraints: endogenous/exogenous interference in a mesh network, half-duplex radios, throughput, and to satisfy application requirements in terms of delay, energy consumption, delivery rate, etc.
The algorithms you will propose will rely on network monitoring data (neighborhood tables, data traffic, interference measurements) and will propose the installation of transmission paths and resources in the network to guarantee the target QoS.
The main expected results are the design of these learning algorithms allowing the calculation of communication schedules in multi-hop networks and the establishment of backup routes in case of transmission failure, their integration in a network controller and a demonstrator.
Funding category: Cifre
PHD Country: France
You are a graduate or future graduate of an engineering school and/or a research master's degree (BAC+5), with a specialization in data science and/or network/telecommunications.
You have the following skills:
You have a very good knowledge of neural network machine learning, and more generally of machine learning techniques.
You have a strong knowledge of networks, preferably wireless.
You have significant experience in python development, especially with Pytorch and/or Tensorflow libraries.
You have experience in C development, ideally in an embedded context.
You are fluent in English. The work done in the framework of this thesis will be the subject of patent applications and publications in international conferences.
You are methodical, curious, autonomous and motivated to tackle an open research topic.
You have good interpersonal skills and know how to communicate on your subjects of interest.
A first experience in the world of research is a plus (R&D internship, scientific paper writing).
- Number of offers available
- Orange Innovation
Where to apply
View or Apply
To help us track our recruitment effort, please indicate in your cover//motivation letter where (vacanciesin.eu) you saw this job posting.