Scholarlink

The Internet of Things (IoT) proliferation has garnered widespread attention due to its versatile applications across manufacturing, commerce, and education industries. Within urban contexts, IoT plays a pivotal role in transforming cities into smart cities (SCs) by offering many applications and services. Nevertheless, these innovative urban solutions carry a significant environmental challenge: air pollution. In this context, the utilization of Deep Learning (DL) and Federated Learning (FL) techniques presents a solution for addressing forecasting challenges, particularly in scenarios involving volatile patterns of air contaminants (ACs) within vast and diverse datasets. The synergy of DL and FL has inspired us to harness the power of DL-based Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) models. These models form the foundation of our novel approach: Long Short-Term Gated Recurrent Unit (LSU). We employ the LSU method for accurate predictions of future air quality, while FL facilitates our model’s secure, decentralized, and distributed training. This study aims to achieve exact future air quality predictions by employing this innovative approach. To gauge the efficacy of the FeLLU framework, we conduct a comparative analysis against other machine learning models, utilizing a range of evaluation metrics.

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

FeLLU: Federated Learning-Based LSU Model for Smart Cities Air Quality Forecasting

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Kalyan Chatterjee,
  • B. Bhavani,
  • Samala Suraj Kumar,
  • Vishali Kancharala,
  • Vaddagoni Saicharan Teja,
  • Parupally Anurag

摘要

The Internet of Things (IoT) proliferation has garnered widespread attention due to its versatile applications across manufacturing, commerce, and education industries. Within urban contexts, IoT plays a pivotal role in transforming cities into smart cities (SCs) by offering many applications and services. Nevertheless, these innovative urban solutions carry a significant environmental challenge: air pollution. In this context, the utilization of Deep Learning (DL) and Federated Learning (FL) techniques presents a solution for addressing forecasting challenges, particularly in scenarios involving volatile patterns of air contaminants (ACs) within vast and diverse datasets. The synergy of DL and FL has inspired us to harness the power of DL-based Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) models. These models form the foundation of our novel approach: Long Short-Term Gated Recurrent Unit (LSU). We employ the LSU method for accurate predictions of future air quality, while FL facilitates our model’s secure, decentralized, and distributed training. This study aims to achieve exact future air quality predictions by employing this innovative approach. To gauge the efficacy of the FeLLU framework, we conduct a comparative analysis against other machine learning models, utilizing a range of evaluation metrics.