Aplicando IA/ML para Gestão Adaptativa de Recursos na Arquitetura Open RAN: Habilitando URLLC em Redes Pós-5G

• Processo: 25/00263-9
• Modalidade de apoio: Bolsas no Brasil - Pós-Doutorado
• Data de Início da vigência: 01 de junho de 2025
• Data de Término da vigência: 31 de maio de 2027
• Área de conhecimento: Ciências Exatas e da Terra - Ciência da Computação - Sistemas de Computação
• Pesquisador responsável: Fabio Luciano Verdi
• Beneficiário: Rohi Tariq
• Instituição Sede: Centro de Ciências em Gestão e Tecnologia (CCGT). Universidade Federal de São Carlos (UFSCAR). Campus de Sorocaba. Sorocaba , SP, Brasil
• Empresa: Universidade Estadual de Campinas (UNICAMP). Faculdade de Engenharia Elétrica e de Computação (FEEC)
• Vinculado ao auxílio: 21/00199-8 - Redes e serviços inteligentes rumo 2030 (SMARTNESS), AP.PCPE
• Assunto(s): Aprendizado computacional Tecnologias 5G Redes de computadores
• Palavra(s)-Chave do Pesquisador: machine learning | Network Slicing | Open RAN | Ultra Reliable Low Latency | 5G | Redes de Computadores

Resumo

Fifth-generation mobile networks (5G) are designed to support diverse use cases, including enhanced mobile broadband (eMBB), massive machine-type communications (mMTC), and ultra-reliable and low-latency communications (URLLC). However, most commercial 5G networks have primarily focused on eMBB, emphasizing higher data rates. As research on mobile networks evolves beyond 5G, the emphasis shifts toward enabling URLLC to support mission-critical applications such as autonomous vehicles, industrial automation, and extended reality.In this context, Network slicing is a key paradigm, allowing multiple virtual networks to operate simultaneously on shared physical infrastructure. While this facilitates diverse applications, the underlying physical infrastructure remains inherently complex, involving heterogeneous topologies, disaggregated components, multi-vendor environments, and cloud-based functionalities.The Open RAN architecture, promoted by the O-RAN Alliance, offers an open, interoperable, virtualized, and intelligent framework to address such complexities. It leverages open interfaces, RAN Intelligent Controllers (Non-RT RIC and near-RT RIC), and the Service Management and Orchestration (SMO) framework to enhance flexibility and scalability. However, integrating URLLC applications into Open RAN, alongside other use cases, poses challenges like real-time decision-making, high-accuracy predictions, and processing vast amounts of data.Artificial intelligence (AI) and machine learning (ML) provide transformative solutions by enabling real-time, adaptive resource management, dynamic optimization of network slices, and efficient handling of large data volumes. This work plan focuses on applying AI/ML to optimize Open RAN architectures for beyond-5G networks, enabling URLLC applications. The methodology involves transitioning from MATLAB and NS-3 simulations to real-world deployments using open-source tools like OpenAirInterface (OAI) and srsRAN, ultimately aiming to develop testbeds and hardware demonstrations. (AU)