A novel Target-DQN model for handover control parameters optimization in ultra-dense cellular networks

Abstract

Efficient mobility management in ultra-dense 5G networks remains a critical challenge, especially in urban environments with high user densities and dynamic conditions. This thesis presents a Deep Q-Network (DQN)-based framework for optimizing handover control parameters (HCPs), including Time-to-Trigger (TTT), Hysteresis (Hys), and A3Offset. The proposed Target-DQN (TDQN) model incorporates a tunable Search Radius (SR) in its action definition to control the action space granularity, enabling scalable multi-parameter optimization. A custom State representation captures key user and network conditions, by means of outage probability, user mobility profile, and cell load. The model is trained using a reinforcement learning approach that leverages real-time user measurements and network states to improve handover performance. Extensive simulations under various SR settings, HCP combinations, and user mobility profiles demonstrate that the proposed model significantly reduces unnecessary handovers and ping-pong events up to 53% and 98% respectively while improving average user throughput by 10% and outage probability by 12% compared to a best practice benchmark recommended by HUAWEI technical documents. The study also investigates the trade-offs between computational complexity and performance, highlighting the model’s adaptability across different optimization granularities and user speeds. These findings suggest that the proposed TDQN framework is a promising solution for dynamic and context-aware mobility optimization in next-generation cellular networks.

Date	12 Aug 2025
Original language	American English
Awarding Institution	École de technologie supérieure
Supervisor	Zbigniew Dziong (Supervisor)