Uploaded June 2026 | Updated September 2026, 3 weeks ago
Decoding RSSI Compression in RFID: Dynamic RCS Modeling and Tag-Intrinsic Power Metrics for Reliable Backscatter Networks
Jia Liu, Yifei Ma, Xingyu Chen, and Haipeng Dai, Nanjing University; He Huang, Soochow University; Zihao Lin and Wei Zheng, Nanjing University; Junzhao Du, Xidian University; Guihai Chen, Nanjing University
Radio Frequency Identification (RFID) is a foundational element of modern IoT and backscatter networks, powering inventory, localization, and battery-free sensing at scale. In this paper, we uncover RSSI compression, a power-dependent bias in reader-measured RSSI, as a critical physical-layer problem that propagates upward in the network stack, degrading MAC-layer collision resolution, network-layer link estimation, and application-layer reliability. Through carefully designed experiments, we trace this distortion to dynamic tag Radar Cross Section (RCS) behavior and introduce two novel physical-layer metrics: Interrogation Threshold Power (ITP), a channel-specific metric for accurate link-quality estimation, and Backscatter Power Index (BPI), a tag-intrinsic, environment-agnostic signature. These metrics provide high-fidelity signal information that higher layers can directly exploit for more robust collision detection, power control, localization and sensing tasks. Finally, an in-situ single-query method further reduces measurement overhead by 99.8%, while cutting channel-estimation error by 64.7%, delivering significant cross-layer performance gains in real-world backscatter networks.
View the full NSDI '26 program at usenix.org/conference/nsdi26/technical-sessions
Decoding RSSI Compression in RFID: Dynamic RCS Modeling and Tag-Intrinsic Power Metrics for Reliable Backscatter Networks
Jia Liu, Yifei Ma, Xingyu Chen, and Haipeng Dai, Nanjing University; He Huang, Soochow University; Zihao Lin and Wei Zheng, Nanjing University; Junzhao Du, Xidian University; Guihai Chen, Nanjing University
Radio Frequency Identification (RFID) is a foundational element of modern IoT and backscatter networks, powering inventory, localization, and battery-free sensing at scale. In this paper, we uncover RSSI compression, a power-dependent bias in reader-measured RSSI, as a critical physical-layer problem that propagates upward in the network stack, degrading MAC-layer collision resolution, network-layer link estimation, and application-layer reliability. Through carefully designed experiments, we trace this distortion to dynamic tag Radar Cross Section (RCS) behavior and introduce two novel physical-layer metrics: Interrogation Threshold Power (ITP), a channel-specific metric for accurate link-quality estimation, and Backscatter Power Index (BPI), a tag-intrinsic, environment-agnostic signature. These metrics provide high-fidelity signal information that higher layers can directly exploit for more robust collision detection, power control, localization and sensing tasks. Finally, an in-situ single-query method further reduces measurement overhead by 99.8%, while cutting channel-estimation error by 64.7%, delivering significant cross-layer performance gains in real-world backscatter networks.
View the full NSDI '26 program at usenix.org/conference/nsdi26/technical-sessions
