Global context and problem statement

Your PhD project will tackle the Forecasting of latencies in cellular networks with Machine Learning.

A forecast of the future latency would allow adapting the network and real-time or interactive applications (e.g. autonomous driving, telerobotics) before the occurrence of an unsuitable latency thanks to proactive actions either to tame the future latency or to endure it. Such forecasts are particularly demanded by the 5G Automotive Association [2] and the telecom industry expects the disaggregation of the base station to facilitate its implementation [1]. Helping latency sensitive applications to use network services offering weak performance guarantees will make low latency more affordable.

Disaggregated base stations [1], supervision tools, transport and application layer softwares, mobile user devices and open data sources provide a vast amount of latency related data.

Machine Learning technics [3] provides the opportunity to accurately forecast the latency of cellular communications, in time and space [4].

Your PhD project, relying on Machine Learning and Artificial Intelligence, will leverage these data to establish how to best forecast the latency of cellular communications.

References :

[1] https://www.o-ran.org/s/O-RAN-Use-Cases-and-Deployment-Scenarios-Whitepaper-February-2020.pdf

[2] https://5gaa.org/news/5gaa-releases-white-paper-on-making-5g-proactive-and-predictive-for-the-automotive-industry/

[3] Boutaba, R., Salahuddin, M. A., Limam, N., Ayoubi, S., Shahriar, N., Estrada-Solano, F., & Caicedo, O. M. (2018). A comprehensive survey on machine learning for networking: evolution, applications and research opportunities. Journal of Internet Services and Applications, 9(1), 16.

[4] Samba, A., Busnel, Y., Blanc, A., Dooze, P., & Simon, G. (2018). Predicting file downloading time in cellular network: Large-Scale analysis of machine learning approaches. Computer Networks, 145, 243-254.

Scientific objective – obstacles and expected results

The scientific objective of this PhD project is to design and assess latency forecasting models applicable to various traffic types (sporadic, high bitrate…) in a variety of network conditions (load, radio configuration…) and for a diversity of usages (mobility, user device…), using the available sources of data.

The main challenge to address are:

The forecastability of end-to-end latency, depending on the time horizon,

Data acquisition and pre-processing, and the design of a unified end-to-end latency forecasting model capable of working with various traffic types, network and usage conditions

About you

• Expected Skills (scientific, technical and personal)
• A good knowledge of mobile networks,
• Methodological competencies in Machine Learning
• Proficiency in Python or R programming languages
• Educational background and diploma (master of science, engineering degree, scientific and technical domains …)
• Ideally a double specialty in Data Science and Networks, validated by and engineering degree or a master of science.
• Otherwise, the candidate should have one of the two specialty and a strong appetency for the other
• Work experience (traineeship, …)
• Ideally a previous traineeship involving Machine Learning or Networks

Additional information

The PhD candidate will work in a team well recognized for its competencies on network automation and machine learning and who is particularly active in open-source projects of the Linux Foundation.

The PhD candidate will benefit from an environment blending research activities with activities aimed at supporting Orange’s business units. This facilitates the émergence of fresh ideas and the development of proof of concepts. The PhD project being the continuation of an ongoing research project i twill benefit from the tools and knowledge previously acquired regarding data collection, preprocessing and first attempts to forecast cellular latencies.

department

Orange 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 Orange Innovation, you will be embedded in the research team “Open and Smart sOlutions for automating Network Services (OSONS)” composed of 15 persons including several PhD candidates at the forefront of innovation and expertise on future networks. The team’s scope covers: contributing to key future network open-source communities such as Magma, ONAP and AcumosAI from the Linux Foundation; supporting Orange’s transformation toward automation and toward the application of machine learning to network automation

Contract

Thesis