Scholarlink

<p>The efficiency of the sixth-generation (6G) communication system is geared toward meeting a variety of user service needs, which has garnered significant attention particularly with the increasing popularity of the Internet of Things (IoT). The massive connectivity of machine-type users, along with ultra-reliable low latency, is critical for dependable communication in IoT scenarios. Achieving these simultaneous service requirements, i.e., massive connectivity and ultra-low latency access—poses a complex challenge within the context of Random Access (RA). To address this critical challenge, we propose a Long Short-Term Memory (LSTM) prediction model to estimate the number of active IoT devices, enabling efficient resource allocation and minimizing access latency. We enhance this model with an attention mechanism (AM), leveraging LSTM’s power to capture long-term dependencies in sequential data while enabling the attention module to focus on key network states and improve decision-making efficiency. This LSTM-AM RA approach is particularly well-suited for the dynamic and diverse environments of integrated terrestrial and non-terrestrial networks, where flexibility and scalability are essential. Numerical results demonstrate that, compared to conventional RA schemes, the LSTM-AM RA framework with attention significantly improves system performance and the number of successful connections. This improvement underscores the feasibility of achieving extensive connectivity and ultra-reliable, low-latency communication in future 6G IoT deployments.</p>

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

LSTM Empowered Random Access with Attention Mechanism for IoT in Integrated Terrestrial and Non-Terrestrial Networks

  • 期刊论文

  • 发表日期：

    2025年8月1日

查看全文
  • Yazeed Alkhrijah,
  • Muhammad Waleed Aftab,
  • Sami Dhahbi,
  • Bouthaina Dammak,
  • Asma A. Alhashmi,
  • Abdulbasit A. Darem

摘要

The efficiency of the sixth-generation (6G) communication system is geared toward meeting a variety of user service needs, which has garnered significant attention particularly with the increasing popularity of the Internet of Things (IoT). The massive connectivity of machine-type users, along with ultra-reliable low latency, is critical for dependable communication in IoT scenarios. Achieving these simultaneous service requirements, i.e., massive connectivity and ultra-low latency access—poses a complex challenge within the context of Random Access (RA). To address this critical challenge, we propose a Long Short-Term Memory (LSTM) prediction model to estimate the number of active IoT devices, enabling efficient resource allocation and minimizing access latency. We enhance this model with an attention mechanism (AM), leveraging LSTM’s power to capture long-term dependencies in sequential data while enabling the attention module to focus on key network states and improve decision-making efficiency. This LSTM-AM RA approach is particularly well-suited for the dynamic and diverse environments of integrated terrestrial and non-terrestrial networks, where flexibility and scalability are essential. Numerical results demonstrate that, compared to conventional RA schemes, the LSTM-AM RA framework with attention significantly improves system performance and the number of successful connections. This improvement underscores the feasibility of achieving extensive connectivity and ultra-reliable, low-latency communication in future 6G IoT deployments.