ROC Seminars – July 15, 2026

In our next ROC seminar, the M2 interns will present their internship topics and current work on Wednesday, 15 July 2026, at 10 AM.

Seminar Location: Amphitheater George Friedmann (access 33, 2^nd floor), 2 Rue Conté, CNAM, Paris, France

1. David Khvedelidze
   • Title: Session-Preserving UPF Scaling with a Customized Open5GS 5G Core
   • Intro: His internship focuses on developing a cloud-native 5G core capable of dynamically adding and removing User Plane Functions (UPFs) without disconnecting active users. His main responsibility is the design and implementation of the core-network scaling workflow. He customized Open5GS, particularly the Session Management Function (SMF), to provide real-time UPF and PDU-session information, administrative UPF draining, and controlled live-session migration. He also developed the Kubernetes-based UPF orchestration and validation components, integrated the system with OpenAirInterface, and ensured that routing and user traffic remain consistent during migration.
2. Tarek Morsi
   • Title: DRL-Based Autonomous Reconfiguration for Cloud-Native 5G Systems
   • Intro: ARM, proposed by Modina et al., is a DRL-based framework for autonomous reconfiguration of network functions. This internship implements the idea on a realistic 5G testbed (srsRAN + Open5GS). The main effort is building a working closed loop on a real testbed: telemetry, anomaly detection, DRL decisions, and Kubernetes actuation all connected through a shared messaging layer.
3. Wahib Kanaan
   • Title: Energy-Efficient Joint Trajectory and RAN Processing Optimization for UAV 5g Base Stations.
   • Intro: Unmanned Aerial Vehicle base stations (UAV BSs) are a flexible way to provide 5G coverage where the ground network is missing, saturated, or unavailable, since a drone can be deployed in minutes and repositioned in real time. Their main constraint is energy. Flying, transmitting, and running the base station software all draw from the same limited battery, so energy efficiency decides how long the drone stays useful. This internship studies a joint management of two levers at once. The trajectory of the UAV drives the radio power, and the processing load of its onboard base station drives the CPU power. Both are controlled together with real software on an Open RAN (O-RAN) framework, and the approach is demonstrated on a real testbed.
4. Luka Tsivilashvili
   • Title: Network digital twins: combination of Nvidia’s SionnaRT and ns-3-rt
   • Intro: His project focuses on network digital twins: combination of Nvidia’s SionnaRT and ns-3-rt that replaces the statistical radio models of the ns-3 network simulator with deterministic ray tracing from NVIDIA Sionna RT on 3D city models, so that real applications can be evaluated over physically accurate 5G links along realistic vehicle routes. Classical network simulators model radio propagation statistically, decoupling link quality from the real environment. SionnaNsRT couples the ns-3 simulator (5G-LENA NR) to NVIDIA Sionna RT, a GPU ray tracer that computes propagation on a 3D city scene built from OpenStreetMap. Real Linux applications cross the simulated 5G link into a network namespace acting as the UE, while user mobility comes from SUMO traffic on the same OSM area, geo-referenced into the scene at centimeter accuracy.
5. Seonho Kim
   • Title: NTN Network Emulation
   • Intro: Non-terrestrial networks use satellites and other aerial platforms to extend 5G and future 6G coverage beyond the reach of terrestrial infrastructure. Because these links introduce higher delay and different loss conditions, their integration with terrestrial networks requires suitable multi-connectivity mechanisms. A simulation platform was therefore developed using ns-3 and 5G-LENA. A functional PDCP layer and a second independent radio path were added, while the RLC behavior was adapted and packet tracking was introduced across the protocol stack. The simulator supports terrestrial, NTN-like, and hybrid scenarios with configurable delay and packet loss, as well as primary-only transmission, packet duplication, and traffic splitting.