Scholarlink

<p>The swift proliferation of Industrial Internet of Things (IIoT) devices and 5G networks presents significant challenges, including the escalating sophistication of cyberattacks (such as ransomware and targeted malware) on resource-limited devices, alongside the elevated energy consumption and latency associated with processing computational tasks on smart terminals. This study introduces a cohesive and streamlined architecture for security and efficiency in Industrial Internet of Things (IIoT) settings facilitated by 5G and Multi-Access Mobile Edge Computing (MEC). The intrusion detection section presents the LATE-IDS system, which utilizes diminutive neural networks integrated with an attention mechanism and energy-aware multi-objective optimization. This approach streamlines the architecture and attains an optimal equilibrium among accuracy, latency, and energy consumption through evolutionary search. This system attains over 97% accuracy and F1-score in identifying known attacks and 74–80% in recognizing unknown (open-set) assaults on battery-operated edge nodes (e.g., Raspberry Pi 4), while decreasing latency by 55–66% and energy consumption by 0.7 mWh per inference. A cluster-aware load balancing method utilizing the Bird Swarm Optimization algorithm is proposed in the resource optimization section for 5G macrocells and small cells, enhancing resource allocation in a multi-MEC and non-orthogonal access (NOMA) environment through collaboration with edge servers and the cloud. The issue is characterized as a restricted, non-convex multi-objective problem, and a global optimum solution is attained by a hybrid intelligence methodology. Experimental findings indicate that this strategy substantially decreases energy usage and delay in terminal activities, while enhancing user experience and processing efficiency.</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

Lightweight Attention-Guided Intrusion Detection and Bird Swarm Energy Optimization for Secure IIoT in 5G MEC

  • 期刊论文

  • 发表日期：

    2026年4月3日

查看全文
  • Jun Lv,
  • Kai Liu

摘要

The swift proliferation of Industrial Internet of Things (IIoT) devices and 5G networks presents significant challenges, including the escalating sophistication of cyberattacks (such as ransomware and targeted malware) on resource-limited devices, alongside the elevated energy consumption and latency associated with processing computational tasks on smart terminals. This study introduces a cohesive and streamlined architecture for security and efficiency in Industrial Internet of Things (IIoT) settings facilitated by 5G and Multi-Access Mobile Edge Computing (MEC). The intrusion detection section presents the LATE-IDS system, which utilizes diminutive neural networks integrated with an attention mechanism and energy-aware multi-objective optimization. This approach streamlines the architecture and attains an optimal equilibrium among accuracy, latency, and energy consumption through evolutionary search. This system attains over 97% accuracy and F1-score in identifying known attacks and 74–80% in recognizing unknown (open-set) assaults on battery-operated edge nodes (e.g., Raspberry Pi 4), while decreasing latency by 55–66% and energy consumption by 0.7 mWh per inference. A cluster-aware load balancing method utilizing the Bird Swarm Optimization algorithm is proposed in the resource optimization section for 5G macrocells and small cells, enhancing resource allocation in a multi-MEC and non-orthogonal access (NOMA) environment through collaboration with edge servers and the cloud. The issue is characterized as a restricted, non-convex multi-objective problem, and a global optimum solution is attained by a hybrid intelligence methodology. Experimental findings indicate that this strategy substantially decreases energy usage and delay in terminal activities, while enhancing user experience and processing efficiency.