無線射頻辨識是一種自動辨識技術，這種技術利用無線電頻率去獲得電子標籤內的資訊。在多個電子標籤的環境中，數電子標籤同時回應讀取器將會成碰撞的問題而危及識別的速度。故此，某些為調解衝突與改善讀取時間的機制是必要的‧本文提出一個新的機制以改善上述問題‧另外，此機制可以被訓練來加快辨識已知的電子標籤‧僅管有一些相似的機制能夠達到同樣的效果，但此機制可以識別未知的電子標籤等罕見狀況。在模擬情況中，本文針對數個電子標籤防碰撞機制進行績效評估‧更精確而言，本文著重在完成讀取所須的時間和訊息數。我們觀察得知本機制能有效減少讀取時間和讀取訊息。在最佳情況下，相較於詢問樹演算法本機制能節省50％的時間;相較於二元樹演算法本機制能節省70％的時間。這些改善不只加快了讀取的速度，也節省了讀取器中之電力。

RFID is an automatic identification technology which exercises the radio frequency to retrieve the information saved in electronic tags. In a multi-tag environment, the collision problem caused by a number of tags simultaneously responding to a reading will jeopardize the speed of identification. Some scheme is necessary to reconcile the conflict and improve the reading time. A novel scheme is proposed in this paper to address this collision problem.In addition, our scheme can be trained to accelerate the identification of a subset of the known tags. Although there are similar schemes capable of achieving this function, our pro-
posal accommodates rare presence of those tags outside the known set. In the simulation, we evaluate the performance of several tag anti-collision protocols. Specifically, we focus on the total reading time and the amount of messages required accomplishing the reading pro-cess. We observe that our scheme can effectively reduce both the reading time and the num-ber of reading messages. In the best case, the proposed scheme saves 50% and 70% time when compared with the query tree algorithm (QTA) and binary tree algorithm (BTA) me-thods respectively. These improvements not only accelerate the reading speed but also save the on-board power in the reader.