PhD "Agentic Network and Flow management in a distributed data architecture for 6G networks" F/M

about the role

Your role is to carry out a PhD thesis on: "Agentic network and Flow management in a distributed data architecture for 6G networks."

· Context

The rise of artificial intelligence (AI) is radically transforming the telecommunications landscape. AI opens the way to innovative applications requiring the integration of LLM techniques associated with networked AI agents (agentic networks). In this context, it becomes necessary to manage massive flows of heterogeneous data to effectively train AI models and process certain services in real time (inference).

In this context, telecom operators face a major challenge: managing, adapting, filtering, "augmenting" the data (data flows) generated by or for these AI agents, in a constantly evolving environment, while ensuring reliable and relevant responses from LLMs and AI agents.

· Objectives

The thesis aims to study and define a distributed and intelligent management of data flows in operator networks, aiming to feed AI agents with formatted data (data preprocessing for learning new AI models) to be able to dynamically adapt to the needs of services mainly using LLMs.

Locks:

There is no technical model for a distributed dynamic management of data flows for AI in mobile networks, while guaranteeing the reliability of the results.

There are technical locks to be overcome such as:

- Real-time processing: Design architectures capable of processing and analyzing in real time massive volumes of data for AI applications and/or generated by AI applications in a wide area mobile network.

- Heterogeneity of resources: Efficiently manage resource allocation in a heterogeneous environment comprising equipment with varied capacities (NFs plane control, NAS, UPFs, routers mobile terminals of the user plane, edge servers, etc.).

- Robustness and resilience: Ensure service continuity and the quality of allocation decisions in the presence of network disruptions, equipment failures or malicious attacks.

- Scalability: Guarantee the performance and efficiency of the proposed system when scaling up, with an increasing number of nodes and AI applications.

- Interoperability: Design interfaces and protocols that enable the seamless integration of the proposed system with existing and emerging network technologies (5G, 6G, SDN, NFV).

- Confidentiality and security: Develop mechanisms that ensure the protection of sensitive data and the security of the resource allocation system, while preserving its performance.