Scholarlink

<p>In modern smart agricultural systems enabled by the Internet of Agriculture Things (IoAT), wireless sensor networks (WSNs) are extensively deployed to enable continuous monitoring of crop conditions and environmental parameters. However, sensor data in agricultural WSNs often face quality challenges such as outliers and missing values, arising from hardware limitations, environmental disturbances, or unreliable communication. Such issues can compromise analytical accuracy and hamper effective agricultural decision-making. To address this challenge, we propose a novel two-phase data healing approach tailored for agricultural sensor data. In the first phase, outliers are detected using an entropy-enhanced k-means clustering technique, which effectively isolates abnormal readings by evaluating both similarity and uncertainty in the sensory data matrix. In the second phase, we apply a Maximum A Posteriori (MAP) estimation technique to recover missing data—whether occurring in complete rows, columns, or randomly distributed entries—by leveraging the underlying structural features of the sensory matrix. This framework ensures robust preprocessing of agricultural data before further analysis or decision-making. We validate the effectiveness of the proposed method using real-world agricultural datasets, demonstrating a high recovery accuracy of 99.39%. The results confirm that our method outperforms existing approaches in terms of accuracy and computational efficiency, making it well-suited for practical IoAT applications. While the study primarily targets agriculture, the proposed framework is broadly applicable to other domains, such as environmental monitoring, healthcare, industrial IoT, and finance, where similar data quality issues also arise.</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

A novel data healing framework for outlier detection and recovery approach in the internet of agriculture things sensor network

  • 期刊论文

  • 发表日期：

    2025年10月3日

查看全文
  • Rashmita Sahu,
  • Priyanka Tripathi

摘要

In modern smart agricultural systems enabled by the Internet of Agriculture Things (IoAT), wireless sensor networks (WSNs) are extensively deployed to enable continuous monitoring of crop conditions and environmental parameters. However, sensor data in agricultural WSNs often face quality challenges such as outliers and missing values, arising from hardware limitations, environmental disturbances, or unreliable communication. Such issues can compromise analytical accuracy and hamper effective agricultural decision-making. To address this challenge, we propose a novel two-phase data healing approach tailored for agricultural sensor data. In the first phase, outliers are detected using an entropy-enhanced k-means clustering technique, which effectively isolates abnormal readings by evaluating both similarity and uncertainty in the sensory data matrix. In the second phase, we apply a Maximum A Posteriori (MAP) estimation technique to recover missing data—whether occurring in complete rows, columns, or randomly distributed entries—by leveraging the underlying structural features of the sensory matrix. This framework ensures robust preprocessing of agricultural data before further analysis or decision-making. We validate the effectiveness of the proposed method using real-world agricultural datasets, demonstrating a high recovery accuracy of 99.39%. The results confirm that our method outperforms existing approaches in terms of accuracy and computational efficiency, making it well-suited for practical IoAT applications. While the study primarily targets agriculture, the proposed framework is broadly applicable to other domains, such as environmental monitoring, healthcare, industrial IoT, and finance, where similar data quality issues also arise.