This paper aims to address the issue of low through-put for users, caused by the random movement of High-Altitude Platform Stations (HAPS) due to winds. The proposed solution involves developing an Equal Clustering (EC) approach that groups users into high-density clusters, ensuring an equal num-ber of users in each cluster while maintaining low complexity. To further enhance the system's throughput performance, we fine-tune the antenna parameters using a Deep Q-Network (DQN) and the results of the EC clustering. To evaluate the effectiveness of the proposed method, we compare it with three Reinforcement Learning (RL)-based approaches and a K-Means clustering-based method. Simulation results indicate that both the EC method and the EC-aided DQN method successfully enhance the Cumulative Distribution Function (CDF) performance of throughput distribution when compared to the RL-based method for both rotation and shift scenarios. Furthermore, the EC-aided DQN method outperforms the K-Means clustering-based method in terms of the CDF of throughput performance.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    A Low-Complexity Clustering-Aided DQN Method for Dynamic Antenna Control in HAPS


    Contributors:


    Publication date :

    2024-06-24


    Size :

    3417150 byte





    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English



    Dynamic Antenna Control for HAPS Using Geometry-based Method in Multi-Cell Configuration

    Yang, Siyuan / Bouazizi, Mondher / Ohtsuki, Tomoaki et al. | IEEE | 2022



    HAPS COORDINATED FLIGHT SYSTEM

    FUJII TAKASHI / OHTA YOSHICHIKA / HOSHINO KENJI | European Patent Office | 2019

    Free access

    HAPS COOPERATIVE FLIGHT SYSTEM

    FUJII TAKAFUMI / OTA YOSHICHIKA / HOSHINO KENJI | European Patent Office | 2021

    Free access

    HAPS OBSERVATORY IN STRATOSPHERE

    Pavlik, Jiri / Novak, Marek / Snízek, Jan et al. | TIBKAT | 2020