The ability to support seamless user mobility has been the Raison D’etre of mobile networks. In recent years, the cellular network industry has evolved significantly, and mobility management is becoming a major challenge. This is due to factors such as densified deployment, stringent Quality of Service (QoS) requirements of diverse use cases like next-generation URLLC, and evolved network architecture. In this thesis, we identify and tackle the pressing challenges of mobility management in emerging cellular networks, which if not addressed, can potentially become the network's Achilles' heel. We address three primary challenges: lack of suitable tools for investigating mobility, radio link failures due to handover failures, and limitations of the existing handover parameters in the standards for mobility management optimization.

To address first challenge, we propose a tri-pronged approach: developing a computationally efficient and realistic mobility simulator, designing and deploying a state-of-the-art experimental testbed called TurboRAN, and introducing AZTEK, an AI-enabled handover parameter optimization framework that can work with limited training data generated either from the simulator, testbed or real network.

To address the second challenge, we propose and evaluate TORIS (Transmit Power Tuning-based Handover Success Rate Improvement Scheme), a novel data-driven solution to reduce inter-frequency handover failures. TORIS consists of an AI-based model to predict handover failures and a heuristic scheme for tuning the transmit power of cells. Unlike conventional methods, TORIS proactively adjusts cell power when a handover failure is anticipated.

To address the third challenge, we propose and analyze a novel parameter called user individual offset (UIO), which considers user-specific behaviors (i.e., speed, direction, service requirement) for value selection. Our results show that UIO can help resolve long-standing challenges in mobility management without tradeoffs between key performance indicators.