Open PhD Student Position

PhD position at INSA Lyon and Spie ICS on the study of anomalies in complex network topologies

  • Subject: Root Cause Analysis in Complex Network Anomalies
  • Dates: 3 years funding, starting date Fall 2024
  • Location: Inria Agora, CITI Lab, INSA Lyon, Campus LyonTech La Doua, 56 Boulevard Niels Bohr, 69603 Villeurbanne, France

We are inviting applications for one PhD student position on computer networks at the CITI laboratory at INSA Lyon, in collaboration with Spie ICS. The position is funded as a CIFRE PhD for three years, starting from Fall 2024 (precise date at the convenience of the candidate). The successful candidate will join the Agora research team at the Inria premises within the Campus LyonTech La Doua, in Lyon, France. The research activity will be carried out under the supervision of Razvan Stanica and a close collaboration with the engineering team from Spie ICS.

The candidate will conduct research on anomaly detection in computer networks. Modern networks present increasing troubleshooting challenges. This is the consequence of several factors, such as: i) integration of more and more usages, types of end-devices, and access technologies; ii) use of multi-vendor equipment and network services; iii) increased virtualization of network functions. In the end, this results in complex and heterogeneous network topologies, which are difficult to monitor and analyze. When a network performance degradation appears, several questions need to be answered. The first question is where the problem initiated. In a complex, multi-vendor, multi- service, multi-technology network, it is not easy to assess this, knowing that the performance issue can be detected in a certain part of the network, but initiate elsewhere. The second question is why the network problem appeared. The degradation in network performance can be the result of a variety of reasons, or even a combination of these factors, for example: an increase in data traffic, an increase in control traffic, traffic being rerouted, etc. The final question to be answered is what produced the problem. For example, an equipment failure, a service failure, or a link failure can result in similar outcomes from a network perspective. In the same line, a misconfigured device, a bug in a network function, or an external attack might produce the same type of consequence. With this general problem in mind, the general objectives of this PhD are the following:

  • Design a machine learning based framework for root cause analysis suitable for multi-vendor, multi-service, multi-technology complex networks.
  • Based on this framework, design network monitoring solutions capable of detecting very early, and ideally anticipating, network anomalies.
  • Implement and test the framework using real datasets and real operational networks.
To that end, the canidate will have access to unique datasets collected in the operational networks of Spie ICS clients, as well as in large-scale cellular networks.

Requirements for this position include:

  • MS or equivalent in Computer Engineering, Computer Science or related areas
  • Strong background in computer networks
  • Previous experience with machine learning techniques
  • Strong algorithm design and software implementation skills
  • Fluency in written and spoken English (French language skills are not mandadory, but they represent a plus)

For applications, please contact Razvan with the following information:

  • Short cover letter indicating the candidate's research interests, achievements to date and vision for the future
  • CV including a detailed list of projects realised by the candidate
  • Master level grades or equivalent information about the courses valdiated
  • Contact details of 1-2 referees