Energy efficiency is crucial for radio frequency identification (RFID) systems as the readers are often battery operated. The main source of the energy wastage is the collision which happens when tags access the communication medium at the same time. Thus, an efficient anti-collision protocol could minimize the energy wastage and prolong the lifetime of the RFID systems. In this regard, EPCGlobal-Class1-Generation2 (EPC-C1G2) protocol is currently being used in the commercial RFID readers to provide fast tag identification through efficient collision arbitration using the Q algorithm. However, this protocol requires a lot of control message overheads for its operation. Thus, a reinforcement learning based anti-collision protocol (RL-DFSA) is proposed to provide better time system efficiency while being energy efficient through the minimization of control message overheads. The proposed RL-DFSA was evaluated through extensive simulations and compared with the variants of EPC-Class 1 Generation 2 algorithms that are currently being used in the commercial readers. The results show conclusively that the proposed RL-DFSA performs identically to the very efficient EPC-C1G2 protocol in terms of time system efficiency but readily outperforms the compared protocol in the number of control message overhead required for the operation.
Read full abstract