Scholarlink

<p>The transmission of infectious diseases in livestock threatens food security, animal welfare, and rural economies. Traditional contact tracing methods are ineffective in large-scale mobile agriculture systems due to limitations in scalability, infrastructure, and privacy. We introduce <i>SafeHerd</i>, a scalable, privacy-preserving framework for tracing epidemic contact in cattle farming. The system employs a hybrid sensing architecture: Bluetooth-enabled “leader” devices perform secure proximity exchanges with other leaders and detect passive RFID/NFC “follower” tags. Proximity events are recorded using ephemeral ECDHE exchanges and ECDSA signatures. Contact logs undergo local statistical filtering based on Poisson models and are then classified using lightweight machine learning algorithms at the farm gateway. Verified exposure tokens are anonymized and submitted to a central authority, which processes them through a layered AI pipeline that includes Isolation Forests, Autoencoders, DBSCAN, and LSTM encoders. Designed for intermittent connectivity and ESP32-class hardware, SafeHerd achieves strong performance in latency, memory usage, and communication overhead. By integrating cryptographically secure IoT sensing with decentralized AI analytics, SafeHerd enables efficient and reliable epidemic monitoring tailored to livestock environments.</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

SafeHerd: a privacy-preserving AI-powered solution for secure epidemic contact tracing in smart agriculture

  • 期刊论文

  • 发表日期：

    2025年10月31日

查看全文
  • Rihab Boussada,
  • Leila Nasraoui,
  • Leila Azouz Saidane

摘要

The transmission of infectious diseases in livestock threatens food security, animal welfare, and rural economies. Traditional contact tracing methods are ineffective in large-scale mobile agriculture systems due to limitations in scalability, infrastructure, and privacy. We introduce SafeHerd, a scalable, privacy-preserving framework for tracing epidemic contact in cattle farming. The system employs a hybrid sensing architecture: Bluetooth-enabled “leader” devices perform secure proximity exchanges with other leaders and detect passive RFID/NFC “follower” tags. Proximity events are recorded using ephemeral ECDHE exchanges and ECDSA signatures. Contact logs undergo local statistical filtering based on Poisson models and are then classified using lightweight machine learning algorithms at the farm gateway. Verified exposure tokens are anonymized and submitted to a central authority, which processes them through a layered AI pipeline that includes Isolation Forests, Autoencoders, DBSCAN, and LSTM encoders. Designed for intermittent connectivity and ESP32-class hardware, SafeHerd achieves strong performance in latency, memory usage, and communication overhead. By integrating cryptographically secure IoT sensing with decentralized AI analytics, SafeHerd enables efficient and reliable epidemic monitoring tailored to livestock environments.