Page d'accueil // SnT // Research // SIGCOM // Projects // LARGOS: Learning-Assisted Optimization for Resource and Security Management in Slicing-Based 5G Networks

LARGOS: Learning-Assisted Optimization for Resource and Security Management in Slicing-Based 5G Networks

Funding Source: Luxembourg National Research Fund (FNR)
Partners: Agency for Science, Technology and Research (A*STAR), the Institute for Infocomm Research (I²R), Singapore
Principal Investigator: Dr. Symeon Chatzinotas
Scientific Advisors: Dr. Sumei Sun (A*STAR); Prof. Björn Ottersten (UL)
Researchers:  Dr. Lei Lei; Thang Xuan Vu
Start Date: 01/09/2018
End Date: 31/08/2021

About the Project

The upcoming 5G system is envisioned to support efficient and flexible resource provision, and provide customized services to meet the service-specific high performance requirements in a variety of user cases. Network slicing has been recently considered for 5G as a key approach to realize this vision. By slicing the common physical infrastructure into multiple virtual logical slices, each network slice can be configured with virtualized network resources to flexibly satisfy the requests by the slice tenant. In practice, to fully cater to 5G system requirements, a number of challenging issues would arise, such as the unprecedented computational complexity and stringent execution-time constraints in online inter/intra-slice resource optimization. Moreover, the security management issues in resource virtualization, physical infrastructure sharing, and the security risks at caching devises, core/edge clouds, and physical layer communications become a serious concern. Resolving these issues calls for novel optimization methods and algorithmic solutions. Due to highly dynamic and complex features in wireless systems, the conventional iterative optimization approaches may have their limits in supporting online real-time network optimization. This motivates the exploration of new avenues in solution development.

Figure 1: Structure of LARGOS

In LARGOS, we will establish whether machine learning can provide a viable alternative to human-engineered algorithmic solutions, or if it can be embedded to the optimization methods to accelerate the optimization process for tackling the difficult resource and security management problems. LARGOS will leverage the power of machine learning and advanced optimization methods, aiming at devising efficient and competitive learning-based solutions for the future slicing-based 5G networks. The key objective of LARGOS is to let the system learn to manage network resources and security, and support online-optimization and online-learning solutions for real-time 5G network intelligent management. As an outcome of LARGOS, a toolset of machine-learning based optimization algorithms or learning-assisted methods over a network-slicing architecture will be developed to provide efficient and high-quality online-optimization solutions for resource and security management.

Two PhD projects in LARGOS have been planned. The University of Luxembourg will act as the Host Institution in Luxembourg while Agency for Science, Technology and Research (A*STAR), Singapore will act as the Collaborating Institution in Singapore. The PhD projects are planned in collaboration between the SIGCOM group at SnT, and the research team of Dr. Sumei Sun at the Institute for Infocomm Research (I²R), A*STAR, Singapore.

Project Partners

  • SIGCOM, SnT, University of Luxembourg, Luxembourg
  • Agency for Science, Technology and Research (A*STAR), the Institute for Infocomm Research (I²R), Singapore

Management and Supervision Team

Funding Details

  • Funding Source: FNR bilateral Singapore programme

Contact