.png){kind=icon}
.png){kind=icon}
Return
Agentic AI-Enabled Adaptive Power Control for Ambient Backscatter Communications
.png){kind=icon}
PRE
AI
Y
H
F
石
T
DOI：10.1109/tcomm.2026.3717019.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Ambient Internet of Things (IoT) has been identified by the 3rd Generation Partnership Project (3GPP) as an enabling technology for large-scale batteryless connectivity in future 6G networks. However, reliable passive backscatter communication remains challenging under limited reader transmit power and dynamic indoor propagation. In this paper, we propose an agentic artificial intelligence (AI)-enabled adaptive power control framework for ambient backscatter communications. The framework adopts a two-timescale structure: a Lyapunov-based inner loop performs real-time reader power adaptation based on tag read ratio and received signal strength indicator (RSSI) feedback, while a local large language model (LLM) periodically analyzes windowed system statistics and updates high-level control targets. The inner controller applies bounded projected updates with a reliability-driven safety term for fast recovery from deep fades and sudden blocking. The outer LLM guidance provides scenario-aware and interpretable parameter adaptation. Experiments on a real radio frequency identification (RFID) testbed under dynamic and static indoor conditions demonstrate that the proposed framework reduces average reader transmit power while maintaining reliable tag reading and regulating RSSI within the desired operating region.
Keywords:
Ambient IoT
backscatter communication
RFID
power control
Lyapunov optimization
agentic AI
large language models
Journal
IF:
8.3
Papers:
1.2W
Citations:
3.6W
Organization
Cited Papers