Deep reinforcement learning for blockchain in industrial IoT: A survey

With the ambitious plans of renewal and expansion of industrialization in many countries, the efficiency, agility and cost savings potentially resulting from the application of industrial Internet of Things (IIoT) are drawing attentions. Although blockchain and machine learning technologies (especially deep learning and deep reinforcement learning) may provide the next promising use case for IIoT, they are working in an adversarial way to some extent. While blockchain facilitates the data collection that is critical for machine learning under the premise of data regulation rules like privacy protection, it may suffer from privacy leakage due to big data analytics with the help of machine learning. To make machine learning and blockchain useful and practical for diversified services in industrial sectors, it is of paramount importance to have a comprehensive understanding of the development of both technologies in the context of IIoT. To this end, in this paper we summarize and analyze the applications of blockchain and machine learning in IIoT from three important aspects, i.e., consensus mechanism, storage and communication. This survey provides a deeper understanding on the security and privacy risks of the key components of a blockchain from the perspective of machine learning, which is useful in the design of practical blockchain solutions for IIoT. In addition, we provide useful guidance for future research in the area by identifying interesting open problems that need to be addressed before large-scale deployments of IIoT applications are practicable.

Automated summary

With the ambitious plans of renewal and expansion of industrialization in many countries, the efficiency, agility, and cost savings potentially resulting from the application of Industrial Internet of Things (IIoT) are drawing attentions. Blockchain and machine learning technologies may provide the next promising use case for IIoT but they are working in an adversarial way to some extent, with a focus on data privacy and security risks.