Internet of Things (IoT) has gained great popularity in various fields including smart warehouse and intelligent manufacturing. As a building block of IoT network, the Radio Frequency IDentification (RFID) technology enables a large and ever-increasing number of physical objects to be monitored across the Internet via tag identification. Efficiently managing massive tags in RFID systems becomes an important research issue for IoT networks. This paper focuses on a fundamental management problem -tag-sorting, which is to (1) put a set S of identified tags into a certain order by informing each tag t ∈ S of a unique integer ∆(t) ∈ {1, 2, • • • , |S|}, and meanwhile (2) keep unidentified tags from receiving any of these integers. For RFID systems, it is critical to solve this problem as quickly as possible in the sense that, once sorted, every identified tag t ∈ S can be manipulated via t's log 2 (|S|)-bit integer significantly shorter than t's 96-bit long tag-ID (log 2 (|S|) 96), boosting efficiency substantially. The existing works of literature, however, fails to solve this problem rapidly, as they accomplish (1) and (2) separately by using aloha-like protocols and Bloom filters, which incur a long communication time far from the optimum. In this paper, we overcome this drawback by proposing a protocol P sort capable of solving the problem fastly. In particular, this protocol is built with a novel data structure and communication scheme to achieve (1) and (2) simultaneously by using a communication time proven to be much less than the stateof-the-art protocols. The simulation results demonstrate the competence of P sort in achieving about 1.4× speedup than the state-of-the-art solutions.INDEX TERMS IoT network, edge server, RFID systems, tag management, tag-sorting, unidentified tag, identified tag.