IEEE_Tr_CNS.pdf

Recent renewed interest in radio frequency identification (RFID) suggests that low-cost passive RFID tags will play a vital role in the Internet of Things (IoT), bridging the gap between the physical and virtual realms. In RFID systems based on dynamic frame-slotted ALOHA (DFSA), it is essential to precisely estimate the number of tags for proper operation of the systems. The challenge with tag quantity estimation is the limited amount of information available for estimation. In particular, a tag estimation is typically performed on a per-frame basis using a single set of statistics collected during a frame. Moreover, the tag population in a reader’s interrogation zone is steadily increasing with the advent of the IoT era. Nevertheless, most estimators perform well even for highly dense tag populations. In this study, we investigate how the tag population affects the performance of tag estimators. In particular, by establishing several strong laws of large numbers (SLLN) results, we show that the estimates always converge to the actual values as the number of tags becomes large. Based on the SLLN results, we propose a novel estimation method where estimates are given as fixed points of contraction mappings. Finally, we examine the key properties of the proposed estimator.

近期对射频识别（radio frequency identification, RFID）的重新关注表明，低成本无源RFID标签将在物联网（Internet of Things, IoT）中发挥关键作用，弥合物理与虚拟领域之间的鸿沟。在基于动态帧时隙ALOHA（dynamic frame-slotted ALOHA, DFSA）的RFID系统中，精确估计标签数量对于系统的正常运行至关重要。标签数量估计的难点在于可用于估计的信息量有限。具体而言，标签估计通常以每帧为单位进行，使用一帧内收集的单一统计数据集。此外，随着物联网（IoT）时代的到来，阅读器询问区域内的标签数量正稳步增长。尽管如此，大多数估计器即使在标签密度极高的情况下也能表现良好。本研究旨在探究标签数量如何影响标签估计器的性能。具体而言，通过建立若干大数定律（strong laws of large numbers, SLLN）的结果，我们证明了当标签数量增大时，估计值始终收敛于实际值。基于这些大数定律（SLLN）结果，我们提出了一种新颖的估计方法，其中估计值由压缩映射的不动点给出。最后，我们分析了所提出估计器的关键特性。