大规模动态RFID系统标签高效检测与识别的方法研究

Radio-Frequency Identification (RFID) technology has brought many practically innovative applications to various fields, from logistic to inventory management. An RFID system allows online monitoring of tagged objects to get their population estimation and identifications. For example, using monitoring results in an RFID-enabled supermarket, a supermarket manager can easily and accurately capture the flow of goods, forecast when to replenish which goods to empty shelves, and relocate wrongly placed ones. Previous approaches concentrate primarily on the static RFID system, involving the participation of all tags in the monitoring event. They fail to differentiate new (or unknown) tags from old (or known) ones and become inefficient in a large-scale and dynamic RFID system consisting of thousands of tags that are constantly changing. The participation of all tags creates redundant information collection from known tags, causing enormous tag collisions and a long delay of monitoring process. Time-efficiency and energy-efficiency will decrease significantly when known tags are the majority in the whole population. Therefore, how to efficiently monitor tags in large-scale and dynamic RFID systems remains to be an open but critical problem. ..This project will propose new efficient approaches for tag monitoring in large-scale and dynamic RFID systems, focusing on detection and identification of unknown/misplaced tags. To this end, we will address three new sub-problems, new population estimation, unknown tag identification, and misplaced tag identification and localization. These three problems are relevant - for instance, the new population estimation can assist the detection of unknown tags and indicate their percentage in the system; an unknown tag to a reader can be a misplaced one showing distinct category information. Solving these problems, our approaches can make contributions as follows: optimize MAC protocol design by adjusting fame size setting in the famed slotted ALOHA protocol, minimize tag collisions by suppressing redundant replies from known tags, allow responses from a subset of tags to save energy, automatically identify misplaced tags for further correct object relocation, and ultimately improve system throughput and reduce overhead. Furthermore, the proposed approaches have the advantage of achieving high accuracy when a dynamic RFID system contains roaming tags or uses mobile readers to facilitate the tag monitoring process, which has been mostly neglected before. The study of the project can open the discussion as how to efficiently collect information from different sets of tags, push forward the accurate RFID localization investigation, and benefit the future RFID system by incorporating mobile tags/readers into the design.

无线射频识别技术(RFID)作为物联网的关键技术，已广泛应用于物流、零售、仓储和医疗管理等动态领域，用来检测与识别附有标签的物品以及估算物品数量。目前的静态研究方法在解决大规模动态RFID系统中不断变化的标签监控问题还面临较大的挑战。传统的静态方法会重复采集已知标签而产生大量冗余信息，导致信号传输严重冲突和延时，执行效率和准确率低，从而严重影响动态RFID系统整体性能。..本项目将首先建立动态RFID系统模型，对动态标签的实时检测、识别和定位等监控问题给出理论的分析，提出高效优化的协议和算法，并通过模拟和实验的方式验证其高效性和准确性。本项目的创新是从静态研究方法过渡到动态研究方法，通过对动态标签的信息采集来高效管理标签和RFID系统,揭示动态标签比例对系统性能影响的规律。本项目的特色是关注RFID系统中的基础研究课题，即标签的数量估计和识别，为动态RFID系统和物联网的设计提供新思路。

无线射频识别技术(RFID)已经广泛应用于物联网领域。本项目针对RFID技术，做了深入详尽的研究。研究内容包括动态RFID模型的建立，动态标签的识别和统计，信息采集及定位管理。对动态RFID模型的研究，项目分别分析了移动标签和移动阅读器对系统性能的影响。项目也提出了在动态的RFID系统中，如何高效的识别和统计标签。包括从物理层的角度，在多阅读器系统中解决此类问题。项目还提出了新的算法在RFID系统中获得准确的传感器信息和位置信息。新的算法利用了完美哈希计算，动态标签分组，和匿名系统中的完整性确认。..项目的成果发布在计算机领域的顶级会议和顶级期刊上，包括7篇会议文章和9篇期刊论文。顶级的会议论文包括IEEE SECON2017，ICNP2016，SECON2016，ICPP2015,INFOCOM2014，IWQoS2014。顶级的期刊包括IEEE/ACM TNET2016, IEEE TMC2017，TCOM2017，TMC2016，TC2015, TPDS2015, TWC2014。本项目实现了真实的RFID系统平台，采集的大量的数据来验证所提出的新算法和系统的高效性。实验的结果发表在16篇高质量的论文中。利用本项目的成果，我们还成功申请了一项国家专利。本项目的科学意义在于对大规模动态RFID 系统进行基础的算法研究。此项目的理论成果比传统方法具有更好的标签识别准确率，更少的采集时间和能量消耗。实验证明了我们提出的算法具有高效性和实用性，并能解决很多物联网系统中的实际问题，如动态物品的分类信息采集、识别和统计。