Promptly pinpointing mobile RFID tags for large-scale Internet-of-Things In recent years, RFID has increasingly been found in our daily life with various applications such as Internet of Things (IoT) and wireless sensor networks. However, considering a large-scale RFIDsystem, collisions occur when multiple tags transmit signals to RFID reader simultaneously, thereby we observe significant degradation of the efficiency of these applications. Besides, in practice, since tags may move into or out of the transmission range of an RFID reader, three different types of tags coexist: dwelling tags, arriving tags, and leaving tags. In this context, it is necessary to design a novel anti-collision protocol which can distinguish them to achieve efficient tag identification. Nevertheless, we observe that existing protocols fail in fully utilizing the differences of the three types of tags. Therefore, this paper proposes a novel protocol, Durable Tree Slotted ALOHA-based protocol (DTSA), to further improve the efficiency obtained in the literature. In general, DTSA distinguishes the three types of tags based on the obtained information from the last identification process. Moreover, DTSA promptly resolves the collisions for the arriving tags and identifies the dwelling tags one-by-one without any collisions. Evaluation results justify that DTSA is more efficient than state-of-the-art protocols.