Autonomic Mobility Management of Heterogeneous Ubiquitous Networks: A Deep Reinforcement Learning Method

As mobile internet rapidly develops and service requirement diversify, the complexity of mobile and handover scenarios is continually increasing. The ubiquity of mobile entities and the continuous development of virtual connectivity, along with fully covered heterogeneous networks, have increased the frequency of both horizontal and vertical handovers. Traditional mobility management schemes mainly solve the problem of horizontal handover, but there are still challenges in achieving efficient vertical handover in heterogeneous ubiquitous networks. Therefore, we propose a heterogeneous ubiquitous network autonomic mobility management framework based on Deep Reinforcement Learning (DRL) and design an Autonomic Mobility Management Scheme (AMMS). Firstly, to achieve efficient network selection and resource optimization, the Multi- Agent Double Deep Q-Network(MA-DDON) is adopted. Then, to further improve the accuracy of handover decision, a Long Short Term Memory (LSTM) network is trained for user trajectory prediction, Finally, the experimental results show that the proposed demonstrates good robustness, significantly improving the efficiency of network handovers and enhancing service availability, thereby meeting the demand for high-quality connectivity in complex and dynamic network environments.