A blockchain-based edge collaborative detection scheme for construction internet of things

Feng Xiong,Cheng Xu,Wei Ren,Rongyue Zheng,Peisong Gong,Yi Ren

DOI: https://doi.org/10.1016/j.autcon.2021.104066

IF: 10.3

2022-02-01

Automation in Construction

Abstract:For smart construction, it is increasingly necessary to develop innovative methods to automatically detect whether construction workers are wearing safety helmets or not. With the development of Internet of Things and object detection in deep learning, it is possible to enable object detection in end devices. In this paper, we propose a coordinated recognition scheme based on YOLOv3 to automatically detect safety helmets at construction sites. Our scheme provides coordinated recognition from multiple devices to obtain comprehensive detection results with multiple viewpoints, which improves the detection accuracy. Besides, to strengthen the security and reliability of detection devices, we propose a peer-to-peer cooperation scheme which is based on smart contract over blockchain. Our scheme ensures that only trusted devices can initiate or participate coordinated recognition tasks. All access and detection records are stored in the blockchain, which is auditable and traceable. Moreover, we propose a fair video sharing mechanism which encourages trusted devices to actively participate in the coordinated recognition tasks. Only participated nodes can access the shared detection videos from others, which forms an alliance of fair sharing with long term interest. The experimental results and analysis justify that our scheme performs well in terms of security and detection accuracy at few expense of a certain startup delay. Our framework can be employed for detecting other objects related to risks in smart construction.

construction & building technology,engineering, civil