Scholarlink

In today’s world, beamforming in 5th Generation (5G) Multiple Input Multiple Output (MIMO) communication networks improves the signal transmission by navigating the radio waves to the particular receivers. However, multiple challenges encountered such as interference, managing hardware complexity and accurate channel estimation. The traditional methods are failed to resolve the above-mentioned difficulties, to overcome these difficulties this research proposes a robust method using Deuling Deep Q-Networks with Probabilistic Attention mechanism (DDQN-PA). Initially, multicellular orientation framework is utilised to focus on state representation for every individual Base Station (BS) and the spatial distribution. Then antenna design framework is used for optimizing the convergence and minimizes interference. Further, the DDQN-PA technique is employed to sort the actions and estimations of Mobile Stations (MSs) within the network which influences the beamforming decisions. This technique enhances the adaptability of dynamic channel conditions by maximizing the Spectral Efficiency (SE) and Energy Efficiency (EE). The proposed DDQN-PA method achieves better results in Spectral Efficiency of 19. 7 Bits per Second per Hertz (Bps/Hz) and Energy Efficiency of 7.1 Megabits per Second Watts (Mbps/W) while compared with the existing Deep Reinforcement Learning (DRL) method.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Deuling Deep Q-Networks with Probabilistic Attention Mechanism Based Beamforming in 5th Generation Networks

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • P. S. Abdul Lateef Haroon,
  • R. Nayana,
  • S. Aishwarya,
  • S. D. Govardhan,
  • V. Sushma

摘要

In today’s world, beamforming in 5th Generation (5G) Multiple Input Multiple Output (MIMO) communication networks improves the signal transmission by navigating the radio waves to the particular receivers. However, multiple challenges encountered such as interference, managing hardware complexity and accurate channel estimation. The traditional methods are failed to resolve the above-mentioned difficulties, to overcome these difficulties this research proposes a robust method using Deuling Deep Q-Networks with Probabilistic Attention mechanism (DDQN-PA). Initially, multicellular orientation framework is utilised to focus on state representation for every individual Base Station (BS) and the spatial distribution. Then antenna design framework is used for optimizing the convergence and minimizes interference. Further, the DDQN-PA technique is employed to sort the actions and estimations of Mobile Stations (MSs) within the network which influences the beamforming decisions. This technique enhances the adaptability of dynamic channel conditions by maximizing the Spectral Efficiency (SE) and Energy Efficiency (EE). The proposed DDQN-PA method achieves better results in Spectral Efficiency of 19. 7 Bits per Second per Hertz (Bps/Hz) and Energy Efficiency of 7.1 Megabits per Second Watts (Mbps/W) while compared with the existing Deep Reinforcement Learning (DRL) method.