An Image Recovery Method for Non-Retransmission Backscatter Communication

Low-power perception technologies, notable for their micro-watt-Ievel energy consumption, offer the capability for years of continuous data sensing and communication without the necessity for battery replacement. These characteristics make them extremely viable for large-scale deployment and position them as a pivotal trend in the future development of the Internet of Things. Low-power video transmission is a critical application. Low-power devices often have weaker signals, which can lead to consecutive packet loss, resulting in severe image defects at the receiver end. To address this, we propose a novel image transmission method at the transmitter side, which transmitting image blocks or rows using a specific sequence, significantly reducing the risk of consecutive block loss under severe channel interference or at low signal-to-noise ratio conditions, thereby en-hancing system stability under continuous packet loss conditions. Therefore, We adjust the deep learning model for real-world scenarios, achieving better image restoration in environments with high packet loss rates. We constructe a prototype based on backscatter communication cameras. Experiments conducte on this platform demonstrate that we can recover the majority of image information, even with a packet loss rate of 75%, by using traditional algorithms or deep learning approaches. This significantly enhances the visibility of low-power visual perception systems.