A Robust and Efficient Anti-Collision Technique for RFID Readers in the Mobile Environments based Machine Learning

RFID is one of the most common wireless techniques utilized in IoT applications. In order to successfully implement and distribute the RFID system over a wide range of industry sectors, a dense enough RFID network is necessary to interfere with RFID tags and readers. To avoid these complications, it is important to properly utilize the available communication resources. In this study, we introduce a set of anti-collision protocols for decentralized training among RFID readers, based on Hybrid Neural Networks (Artificial Neural Network and Support Vector Machine), that can predict collisions and guarantee optimal resource distribution by factoring in the motion of tags and clients. Our anti-collision model is evaluated for a number of healthcare mobility use cases to guarantee that the collected data meets the necessary throughput, latency, overloading, security, and energy consumption benchmarks. This reduces the potential for network congestion and sluggish connections by eliminating the need for readers to share control information during the application phase. The simulation results based on the number of readers and resource usage demonstrate the vigorous and efficiency of the anti-collision model. Using the selected model, several readers may efficiently query tags at the same time, making the most of available frequency and time.