An RFID Anti-Collision Algorithm Integrated with LSTM
Zilin Li, Zhihua Li
Abstract
Conventional anti-collision protocols in RFID (Radio-Frequency Identification) technology grapple with multiple challenges including frequent tag collisions, diminished recognition efficacy, and compromised system reliability. To address these problems, this paper proposes a new RFID anti-collision algorithm, Dynamic Frame Slotted ALOHA based on Tag Grouping and Long Short Term Memory (D-G-MFSA), by integrating LSTM into the existing ALOHA algorithm. D-G-MFSA treats the number of tags within the reader's identification range as a time series for real-time tag prediction, dynamically adjusts the frame length, and groups tags when the number is high. Experimental results indicate that compared to traditional tag anti-collision algorithms, D-G-MFSA demonstrates efficient throughput in scenarios with a limited number of tags, promotes system stability with a high number of tags, and possesses advantages such as simple principles and high reading efficiency.