Banca de DEFESA: CARLOS ANTONIO DE LIMA FILHO

Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.
STUDENT : CARLOS ANTONIO DE LIMA FILHO
DATE: 01/08/2025
TIME: 14:00
LOCAL: Remota via Google Meet
TITLE:

Resource Allocation in Open RAN Networks: An Elasticity Approach for Network Slicing

KEY WORDS:

Network slicing; open RAN; 5G; xApp

PAGES: 60
BIG AREA: Engenharias
AREA: Engenharia Elétrica
SUBÁREA: Telecomunicações
SPECIALTY: Sistemas de Telecomunicações
SUMMARY:

The fifth generation of mobile systems (5G) was designed to meet a wide range of use cases that demand flexibility and efficiency in their configurations to ensure adequate quality of service. The concept of network slicing allows the creation of logical networks (slices) over a shared infrastructure while promoting the characteristics of elasticity, isolation, and customization, which are essential to meet 5G use cases. The efficient allocation of radio resources for each slice is a challenge, in which, depending on traffic conditions, problems related to reduced transmission rate and increased delay may occur. Another facet of the 5G network is the development of solutions for Open Radio Access Network (Open RAN) networks, which allow the implementation of customized algorithms for dynamic optimization of RAN resources, called xApps. The great opportunity is that xApps can be implemented even by those who do not have access to the RAN source code. This master's thesis proposes developing an open RAN solution focused on the dynamic bandwidth allocation for 5G slices. The proof-of-concept methodology uses the O-RAN Alliance's Open RAN architecture and the OpenAirInterface 5G network, enabling commercial 5G terminals through software-defined radio platforms. The fixed allocation of RAN resources is presented as the reference slice scheduling algorithm, called Fixed Slice Scheduling (FSS). The FSS investigation shows that the bandwidth of each slice can be defined by an xApp, preserving slice isolation, even when a RAN is shared by more than one slice. The work proposes the implementation of GBSS (Guaranteed Bandwidth Slice Scheduling), a version of FSS with the definition of a shared bandwidth and the guarantee of a minimum bandwidth for each slice. Finally, the work proposes RLSS, which exploits reinforcement learning to refine the fixed bandwidth allocation solution based on KPIs provided by the RAN.

COMMITTEE MEMBERS:
Presidente - 1412682 - VICENTE ANGELO DE SOUSA JUNIOR
Interno - 2524053 - ANTONIO LUIZ PEREIRA DE SIQUEIRA CAMPOS
Externo ao Programa - 1699087 - AUGUSTO JOSE VENANCIO NETO - UFRNExterno à Instituição - ÁLVARO AUGUSTO MACHADO DE MEDEIROS - UFJF