Distributed Optimization and Machine Learning for for 5G and 6G Virtualized Wireless Networks

Research / Pillars & Groups / Communications & Artificial Intelligence / SNS / PhD Topics / Distributed Optimization and Machine Learning for for 5G and 6G Virtualized Wireless Networks
17 January. 2021
The 5G mobile network architecture is already consolidated in terms of network components, technologies and interfaces. However, this potential cannot be fully harnessed without the appropriate algorithmic innovations and data-driven network automation that will permit full exploitation, global management and end-to-end integration of all the heterogeneous network components and resources. Substantial work is still required to meet the IMT-2020 objectives set for the 5G network performance and fully harness the multitude of technological capabilities offered by 5G with minimum human intervention (e.g., through intelligent network automation and control). Network automation has been investigated since LTE due to its significant impact on reduction of operators’ operational expenditure (OPEX). Different features of network automation have been discussed and studied over the years; however, with recent advances in machine learning standardization bodies are considering a more structured use of learning technique in network automation. In this project we will focus on MEC-empowered service provisioning, and end-to-end network slicing, all integrated and jointly orchestrated by forward-looking data-driven analytics-powered network control and automation. The project will be aligned with considered features in 3GPP release 16 and 17. The successful candidate will investigate the effects of different machine learning approaches on the performance of automated networks with different objectives and primary applications. This includes considering localized automation for “cognitive self-management”, and network level automation for policy-based management.