AbstractWith the ultra-dense small cell deployment and rapid mobility characteristics in 5G, the traditional mobility management method employed in LTE lead to various mobility-related problems, such as frequent handovers and handover failures. On the other hand, the maintenance and operation cost of mobility management increases due to the increasing node density. Mobility management control must have adaptive and predictive features in its core design to overcome these challenges. This thesis aims to develop a self-organising intelligent mobility management based on artificial intelligence (AI) approach for 5G heterogeneous networks.
First, two intelligent handover triggering mechanisms are developed to execute the handover process at the precise moment. The first developed method is based on Q-learning frameworks and subtractive clustering techniques. The input metrics are first converted into state vectors by subtractive clustering, which can improve the efficiency and effectiveness of the training process. Then, the Q-learning framework learns the optimal handover triggering policy from the environment. The trained Q table is deployed on the user equipment (UE) to trigger the handover process. The second developed method integrates the advantages of subtractive clustering and Q-learning framework into the conventional fuzzy logic-based handover algorithm (FLHA). Subtractive clustering is first adopted to generate a membership function (MF) for the FLHA to enable FLHA with the feature of self-configuration. Subsequently, Q-learning is utilised to learn the optimal handover policy from the environment as fuzzy rules that empower the FLHA with a self-optimisation function. The FLHA with self organisation networks (SON) functionality also overcomes the limitations of the conventional FLHA, which relies heavily on professional experience for design.
Second, an optimal handover target selection scheme is developed that integrates the advantages of both fuzzy logic and multiple attribute decision (MADM) algorithms to ensure that the handover connection be switched to the optimal neighbouring base station (BS). To further improve the performance of the proposed scheme, this paper also employs the subtractive clustering technique by using historical data to define the optimal membership functions within the fuzzy system.
Finally, two intelligent one-stage mobility control approaches can not only trigger the handover process accurately, but also select an optimal neighbouring base station (BS) as the handover target. Both methods are based on reinforcement learning with function approximation to autonomously learn an optimal handover control policy by interacting with the environment. The linear function approximator – tile coding and nonlinear function approximator – neural networks are implemented to approximate the value function. The final approximated value function is used to control mobility management.
The combined novel methods and algorithms in this project are expected to enhance the robustness of UE mobility for near-optimal 5G operation. In addition, the algorithms developed in this thesis will have plug-and-play capabilities with the ability to self-optimise and self-configure for easy deployment and maintenance.
|Date of Award||Jul 2022|
|Supervisor||C.F. Kwong (Supervisor) & Jing Wang (Supervisor)|
- reinforcement learning
- mobility management
- self organisation networks
- mobile networks