An algorithm based on Q-learning for solving frequency assignment in RFID systems

RFID (radio frequency identification) systems are used in a variety of applications, and minimizing the number and frequency of reader collision is absolutely essential for RFID systems to function properly. An algorithm that finds dynamic solutions to the reader collision problem in RFID systems is presented. The algorithm is based on a type of reinforcement learning called Q-learning, which is used to determine to frequency and time assignment. The results of the simulations have shown that the techniques used by the algorithm can effectively minimize the effects of the reader collision problem, and the algorithm prefers frequencies over time slots.