Scholarlink

<p>Edge computing acts as a backbone for Internet of Things (IoT) systems to mitigate solutions such as higher latency, low bandwidth utilization, and centralized bottlenecks. Thus, the given paper introduces a novel IoT-based framework equipped with fuzzy logic (FL) and federated learning for decision-making in an edge computing environment. FL ensures that the framework allows for adaptable decision-making in the face of uncertain and imprecise conditions in IoT, thereby improving responsiveness and predictive accuracy. This method preserves data privacy and allows model training to occur without centralized, up-to-date data on the central server, which can be a significant advantage. The performance of the proposed framework is assessed and validated with existing recent studies based on metrics such as accuracy, latency (sec), energy consumption (J), and resource utilization. The results show significant advancements in decreasing latency, increasing energy efficiency, and improving classification accuracy, proving that the proposed framework is applicable and effective for heterogeneous IoT scenarios such as smart cities, industrial automation, and healthcare.</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 IoT-based framework employing fuzzy logic and federated learning for decentralized decision-making

  • 期刊论文

  • 发表日期：

    2025年7月30日

查看全文
  • Vijit Srivastava,
  • Shashi Raj K,
  • Vikas Lamba,
  • Vageesha Salanki Mathada,
  • Chetan Bulla,
  • Neha Gupta,
  • P. Veeramanikandan

摘要

Edge computing acts as a backbone for Internet of Things (IoT) systems to mitigate solutions such as higher latency, low bandwidth utilization, and centralized bottlenecks. Thus, the given paper introduces a novel IoT-based framework equipped with fuzzy logic (FL) and federated learning for decision-making in an edge computing environment. FL ensures that the framework allows for adaptable decision-making in the face of uncertain and imprecise conditions in IoT, thereby improving responsiveness and predictive accuracy. This method preserves data privacy and allows model training to occur without centralized, up-to-date data on the central server, which can be a significant advantage. The performance of the proposed framework is assessed and validated with existing recent studies based on metrics such as accuracy, latency (sec), energy consumption (J), and resource utilization. The results show significant advancements in decreasing latency, increasing energy efficiency, and improving classification accuracy, proving that the proposed framework is applicable and effective for heterogeneous IoT scenarios such as smart cities, industrial automation, and healthcare.