RFID Anti-Collision Algorithm Based on Deep Learning

This paper proposes an RFID tag anti-collision algorithm based on graph neural networks. RFID technology, known for its non-contact and bidirectional data communication features, is widely used in fields such as telecommunication. However, in high-density tag scenarios, the problem of tag collisions can lead to reduced system throughput and recognition efficiency. To address this issue, this paper presents the LTTE algorithm, which uses VOGT algorithm input values and applies LSTM and Transformer models for packet prediction and estimation. When the number of packets is large, Tag-Grouping technology is used to group them, improving system throughput and reducing time slot consumption. Experimental results show that compared to traditional methods, the LTTE algorithm demonstrates better adaptability in complex situations, simplifies the training process, and eliminates the need to train a neural network for each frame length, significantly enhancing RFID system performance.