Towards Privacy Protection: A Non-Invasive Human Detection Method via Edge-Cloud Collaboration

Traditional human detection technologies, such as those based on computer vision or wearable devices, often require the storage and analysis of vast amounts of personal data in the cloud, which increases the risk of data breaches. Therefore, the rapid processing of large-scale detection data while protecting user privacy has become a critical issue that needs urgent resolution. The approaches based on wireless sensing will achieve privacy protection, non-contact perception, massively scalable, and long-range detection. In this paper, we propose an innovative solution that combines data collection from terminal devices, real-time processing by edge computing, and deep analysis capabilities of cloud computing, to realize efficient and privacy-friendly human detection. First, we propose a non-invasive human detection method based on edge-cloud collaboration, integrating wireless sensing, data transmission, and processing within edge nodes, and utilizing drones equipped with edge nodes to process a large amount of wireless sensing data in real-time. Secondly, we designed a human presence prediction module based on a lightweight convolutional neural network to analyze signal data and achieve non-invasive human detection. Finally, with commodity software defined radio devices, we implement a non-invasive human detection prototype that can accurately differentiate multiple states of targets while protecting their privacy. Experimental results indicate that the proposed solution effectively reduces the risk of privacy leakage while maintaining a high detection accuracy. This provides strong support for the application of non-invasive personnel detection technology in the aspect of privacy protection.