Scholarlink

<p>In the 6G network, it is crucial to have edge intelligence, which is the combination of edge computing and the Radio Access Network (RAN) with Artificial Intelligence (AI) for efficient and intelligent applications. The most critical challenge faced by 6G-enabled IoT systems is spectrum resource allocation, which requires an effective solution. There is a significant requirement for an efficient allocation strategy to handle the massive data generated by Internet of Things (IoT) devices. In the Ultra-Massive Multiple-Input Multiple-Output (UM-MIMO) network, the major aspects of resource management are computation power, spectrum, and resource allocation. Interference can be mitigated through effective power allocation along with energy efficiency optimization, which is considered to be a complex task, especially in a high-traffic environment. To manage and support diverse user requirements, spectrum allocation based on the available bandwidth is necessary. Hence, this paper presents a novel resource scheduling policy for managing resource allocation and avoiding overlapping interference in 6G-enabled IoT networks. It adopts a Reinforcement Learning (RL)-based approach to improve network throughput and prevent overlapping interference. The proposed method utilizes the Deep Adaptive and Attentive Multi-agent Actor-Critic model (DA-AMAC), which improves the efficiency of the network. The proposed resource management approach is suitable for industrial automation and intelligent transportation systems. As a further improvement, the proposed model undergoes parameter tuning based on the Levy-influenced Gooseneck Barnacle Optimization (LGBO) algorithm, thereby enhancing the resource allocation process. The presented edge intelligence-based resource scheduling policy can be applied to smart IoT applications as it provides enhanced network efficiency with reduced implementation costs. Finally, the model is implemented and evaluated against different resource allocation models, and its performance is assessed using various evaluation metrics.</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

An Edge Intelligence Framework for Resource Scheduling Policy by Deep Adaptive and Attentive Multi-agent Actor-Critic Model in Ultra Massive MIMO System for 6G Internet of Things

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月29日

查看全文
  • Roqia Rateb,
  • Jafar Ahmad Alzubi,
  • Srinivas Vadali,
  • Muthusamy Palani Rajakumar,
  • Chung Gu Kang,
  • Faheem Khan

摘要

In the 6G network, it is crucial to have edge intelligence, which is the combination of edge computing and the Radio Access Network (RAN) with Artificial Intelligence (AI) for efficient and intelligent applications. The most critical challenge faced by 6G-enabled IoT systems is spectrum resource allocation, which requires an effective solution. There is a significant requirement for an efficient allocation strategy to handle the massive data generated by Internet of Things (IoT) devices. In the Ultra-Massive Multiple-Input Multiple-Output (UM-MIMO) network, the major aspects of resource management are computation power, spectrum, and resource allocation. Interference can be mitigated through effective power allocation along with energy efficiency optimization, which is considered to be a complex task, especially in a high-traffic environment. To manage and support diverse user requirements, spectrum allocation based on the available bandwidth is necessary. Hence, this paper presents a novel resource scheduling policy for managing resource allocation and avoiding overlapping interference in 6G-enabled IoT networks. It adopts a Reinforcement Learning (RL)-based approach to improve network throughput and prevent overlapping interference. The proposed method utilizes the Deep Adaptive and Attentive Multi-agent Actor-Critic model (DA-AMAC), which improves the efficiency of the network. The proposed resource management approach is suitable for industrial automation and intelligent transportation systems. As a further improvement, the proposed model undergoes parameter tuning based on the Levy-influenced Gooseneck Barnacle Optimization (LGBO) algorithm, thereby enhancing the resource allocation process. The presented edge intelligence-based resource scheduling policy can be applied to smart IoT applications as it provides enhanced network efficiency with reduced implementation costs. Finally, the model is implemented and evaluated against different resource allocation models, and its performance is assessed using various evaluation metrics.