Banca de DEFESA: LUCAS ISMAEL CAMPOS MEDEIROS

Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.
STUDENT : LUCAS ISMAEL CAMPOS MEDEIROS
DATE: 31/07/2026
TIME: 09:00
LOCAL: Link da videochamada: https://meet.google.com/ybt-mgqk-eod
TITLE:

Traffic Management in Open RAN Networks with Multi-Connectivity Using Machine Learning

KEY WORDS:

Open RAN, Reinforcement Learning, Dual Connectivity, ns-3, 5G.

PAGES: 50
BIG AREA: Engenharias
AREA: Engenharia Elétrica
SUBÁREA: Telecomunicações
SPECIALTY: Sistemas de Telecomunicações
SUMMARY:

Simultaneously supporting services with distinct requirements, such as Internet browsing and Extended Reality (XR) applications, poses significant challenges for resource management in mobile networks, particularly in dynamic scenarios associated with tourism and smart cities. Conventional Multi-Connectivity approaches, based on static traffic-splitting policies between the Master Node and the Secondary Node, have proven inadequate in coping with demand fluctuations, often resulting in resource underutilization or degradation of the Quality of Service (QoS) of priority applications. This work proposes a dynamic traffic management approach based on the Open RAN architecture, leveraging a Multi-Armed Bandit (MAB) Reinforcement Learning algorithm implemented as an xApp within the Near-Real Time RAN Intelligent Controller (Near-RT RIC). The proof-of-concept was developed on a prototyping platform that integrates the ns-3 network simulator with a real implementation of the Near-RT RIC from the O-RAN Software Community, establishing communication through the E2 interface and the E2SM-KPM and E2SM-RC service models. The results demonstrate that the proposed solution can dynamically adapt traffic distribution according to the network's operational objectives, allowing operators to configure different prioritization policies. As a result, the system can either enhance the Quality of Experience (QoE) of XR users by maximizing their satisfaction or prioritize the performance of enhanced Mobile Broadband (eMBB) users while maintaining the desired QoS levels for each operational scenario. The conducted investigation supports the adoption of Reinforcement Learning as an enabling technology for adaptive automation and zero-touch management of mobile networks. In addition to facilitating the efficient integration of multi-RAT solutions within the 5G ecosystem, the proposed approach contributes to the evolution of future 6G networks, which are expected to provide higher levels of autonomy, intelligence, and adaptability to user and application demands.

COMMITTEE MEMBERS:
Presidente - 1412682 - VICENTE ANGELO DE SOUSA JUNIOR
Interno - 2524053 - ANTONIO LUIZ PEREIRA DE SIQUEIRA CAMPOS
Externo à Instituição - EDELBERTO FRANCO SILVA - UFJF