With the recent development of Internet of Things (IoT) in the next generation cyber-physical system (CPS) such as autonomous driving, there is a significant requirement of big data analysis with high accuracy and low latency. For efficient big data analysis, deep learning (DL) supports strong analytic capability; it has been applied at the cloud and edge layers by extensive research to provide accurate data analysis at low latency. However, existing researches failed to address certain challenges, such as centralized control, adversarial attacks, security, and privacy. To this end, we propose DeepBlockIoTNet, a secure DL approach with blockchain for the IoT network wherein the DL operation is carried out among the edge nodes at the edge layer in a decentralized, secure manner. The blockchain provides a secure DL operation and removes the control from a centralized authority. The experimental evaluation demonstrates that the proposed approach supports higher accuracy.

中文翻译：

---

基于区块链的下一代工业网络物理系统网络安全深度学习方法


随着物联网（IoT）在自动驾驶等下一代网络物理系统（CPS）中的发展，对高精度、低延迟的大数据分析提出了显着的要求。为了高效的大数据分析，深度学习（DL）支持强大的分析能力；通过广泛的研究，它已应用于云和边缘层，以低延迟提供准确的数据分析。然而，现有的研究未能解决某些挑战，例如中心化控制、对抗性攻击、安全和隐私。为此，我们提出了 DeepBlockIoTNet，一种用于物联网网络的区块链安全深度学习方法，其中深度学习操作以去中心化、安全的方式在边缘层的边缘节点之间进行。区块链提供了安全的深度学习操作，并消除了中心化机构的控制。实验评估表明所提出的方法支持更高的准确性。