The Doctoral School in Science and Engineering is happy to invite you to Fatemeh KAVEHMADAVANI’s defence entitled
Artificial Intelligence (AI)-enabled Smart Radio Environments for 6G Wireless Networks
Supervisor: Prof. Symeon CHATZINOTAS
The rapid evolution from fifth-generation (5G) to sixth-generation (6G) wireless networks marks a significant advancement in communication technologies, aiming to support a wide range of services with enhanced performance requirements. Key use cases, such as enhanced mobile broadband (eMBB) and ultra-reliable and low-latency communications (uRLLC), demand robust and flexible network architectures. The Open Radio Access Network (O-RAN) Alliance introduces a transformative approach by advocating for disaggregated RAN functionality using open interface specifications. Within this framework, traffic steering (TS) and user association (UA) become crucial use cases, enabling the efficient management of network resources and the optimization of service delivery. Given this context, the dissertation focuses on designing and developing intelligent TS and UA frameworks for multi-traffic scenarios, leveraging innovative 5G technologies and standardization to enhance network performance and adaptability in dynamic environments by merging machine learning (ML) models. Specifically, the study delves into four key aspects: 1) A TS scheme to jointly allocate heterogeneous network resources in the presence of known dynamic traffic demand and fixed numerology; 2) A Long Short-Term Memory (LSTM)-based traffic prediction strategy for downlink eMBB and uRLLC coexistence within orthogonal frequency division multiple access (OFDMA)-based Open RAN architecture, considering unknown dynamic traffic demands and slice isolation; 3) A centralized Deep Reinforcement Learning (DRL)-based TS scheme for dynamic environments managing diverse services, utilizing slice awareness techniques and flexible numerologies; and 4) A hierarchical optimization-based intelligent UA, congestion control, and resource scheduling scheme, aligned with the 7.2x functional split (FS) in Open RAN architecture, assuming non-orthogonality between radio units (RUs). This thesis contributes to the advancement of adaptive, intelligent solutions for next-generation wireless networks, enabling enhanced performance, scalability, and adaptability in dynamic communication environments.