The i3S laboratory is one of the largest research laboratories in the field of information and communication sciences and technologies on the Côte d'Azur, and it was among the first to establish itself at the Sophia Antipolis science and technology park.
It brings together nearly 300 people, including approximately 80 faculty members—most of whom come from three UniCA (Université Côte d'Azur) entities: Polytech Nice Sophia, the EUR DS4H, and the IUT Côte d'Azur.
The laboratory also includes 20
CNRS researchers and 10 Inria researchers, as well as around 20 technical and administrative staff members.
The workforce is further complemented by over 100 doctoral students, about a dozen post-doctoral researchers, and 60 interns from master's programs or engineering schools.
Affiliated with
CNRS Computer Science, the research themes at the i3S laboratory cover a broad spectrum, spanning areas defined by sections 27 ("Computer Science") and 61 ("Computer Engineering, Automation, and Signal Processing") of the CNU (National Council of Universities), as well as sections 6 and 7 of the CoNRS (National Committee for Scientific Research).
The laboratory is organized into four main departments: - The "Communications, Networks, and Embedded and Distributed Systems (COMRED)" division consists of two joint i3S-Inria research teams (Kairos and Coati) and two research groups: Scale and SigNet.
- The "Discrete Models for Complex Systems (MDSC)" division consists of the MC3 and C&A research groups.
- The "Signal, Images, and Systems (SIS)" division consists of the Signal, SigNet, MediaCoding, and Design research groups, the joint i3S-Inria research team Morpheme, and the Robotics and Autonomous Systems group.
- The
SPARKS cluster comprises three i3S-Inria joint research teams (EPCs)—Wimmics, Maasai, and Marianne—as well as the
PRIME eXplAIn team.
The candidate will be hosted jointly within the Coati joint research team (EPC) and the SigNet team.
Context and Problem Statement Telecommunication infrastructures (fixed and mobile access networks, backhaul networks) are critical components of modern socio-economic functioning.
However, as highlighted by
ARCEP in its note on network resilience [6], these physical infrastructures are highly exposed to the direct and indirect impacts of climate change.
Extreme weather events (heatwaves, storms, floods, wildfires) as well as long-term trend shifts (rising average temperatures, sea level rise) degrade service quality, increase failure rates, and threaten to weaken isolated areas [1, 2].
While initial scientific research and exploratory studies jointly conducted by the SigNet and Coati teams have established empirical correlations between weather conditions and network incidents on a small scale, several major scientific roadblocks remain to generalize these approaches to a national and predictive scale: 1. Spatio-temporal heterogeneity and granularity of data: Crossing complex time series (weather metrics,
RIPE Atlas active/passive probes, ARCEP/operator incident reports, physical network topology) requires processing noisy, incomplete, and asynchronous data.
2.Causality vs. Correlation: Distinguishing failures directly caused by climate factors from those resulting from cascading failures (e.g., power grid outages, traffic saturation during crises).
3. Prospective uncertainty: Anticipating the vulnerability of future network generations (5G standalone, 6G, fiber densification) under various
IPCC climate scenarios for the 2030–2050 horizon.
Scientific Objectives of the Thesis The main objective of this thesis is to build a systematic, automated, data-driven methodology to assess, predict, and enhance the resilience of telecommunication networks in the face of climate hazards.
The research will focus on the following axes: 1. Data Aggregation, Coupling, and Advanced Mapping - Extending the cross-analysis between meteorological time series (Météo-France / Copernicus), incident data (ARCEP), and fixed/mobile access quality (RIPE Atlas, crowdsourced data) across the entire French territory.
- Developing standardized metrics to measure the weather-climate vulnerability index of a piece of equipment or a network node.
2. Explanatory Modeling for Causality Detection - Designing statistical models capable of identifying root causes of incidents and detecting cascading failure propagation (e.g., the impact of a heatwave on cooling systems in cell sites or optical distribution frames / NROs).
- Accounting for cross-dependencies with other critical infrastructures (specifically the
ENEDIS/RTE power distribution network).
3. Prospective Inference and Resilience Optimization Strategies - Simulating network behavior under various future extreme climate scenarios (IPCC / Météo-France projections).
- Proposing redeployment strategies, hardware hardening, or emergency dynamic routing to minimize the risk of global disconnection during extreme events.
Scientific Roadblocks and Expected Contributions - Scaling and automation: Moving from a local exploratory analysis (departmental scale) to an automated data processing pipeline for high-volume streams at a national scale.
- Systemic resilience: Providing a unified theoretical framework combining physical infrastructure characteristics (material exposure) with the dynamic behavior of data traffic flows.
- Public and industrial decision support: Delivering a risk assessment tool that regulators (ARCEP,
ANSSI) and operators (Orange, SFR, Bouygues Telecom, Free, Axione) can use to guide network hardening investments.
Candidate Profile - Master's degree or Engineering degree in Computer Science, Data Science, or Telecommunications.
- Strong skills in data analysis, machine learning / deep learning, and time series analysis.
- Good understanding of telecommunication networks and Internet protocols.
- Keen interest in interdisciplinary challenges (computer science, networks, climate, environment).
Bibliographic References [1] Padmanabhan, R., et al.
"Residential links under the weather." Proc.
of ACM SIGCOMM, 2019. [2] Durairajan, R., Barford, C., & Barford, P.
"Lights out: Climate change risk to internet infrastructure." Proc.
of ANRW, 2018. [3] Météo-France — Open data platform (meteo.data.gouv.fr).
[4]
RIPE NCC —
RIPE Atlas Infrastructure(atlas.ripe.net).
[5]
ARCEP — Incident tracking and mobile network status (arcep-dev.github.io/siteshs/).
[6]
ARCEP — Networks of the future: The resilience of electronic communications networks, Summary Note.