【作者】 吴楠；

【导师】 钱志鸿；

【作者基本信息】 吉林大学 ， 通信与信息系统， 2014， 硕士

【摘要】 作为新时代信息科技的产物，射频识别技术(Radio Frequency Identification, RFID)具有巨大的应用价值。它以无线电波技术为基础，利用空间电磁耦合实现无线双向数据通信和对物品的自动识别。RFID技术的发展得益于多门技术的成熟，涉及计算机、无线数字通信、电磁学、密码学等众多科学领域，其本身具有穿透性强、读取数据方便快捷、自动化程度高、易于操作等突出特点。迄今为止，RFID的应用范围已扩展到工业生产和日常生活的方方面面，为人们带来了极大方便。近年来，人们对于物联网（Internet of Things, IOT）越来越认可，对这一技术的研究也越来越趋于深入和广泛。在物联网整个体系架构中，RFID技术是其感知层的关键技术之一。在未来物联网的应用环境下，将充分利用RFID技术采集和传输底层信息，其优势之一就是多目标识别，即快速准确地识别可读范围内的大量标签。但由于物联网特有的海量数据环境，多个标签在争用同一信道传输数据时不可避免地导致了数据碰撞问题的出现，影响系统数据传输的完整性和效率。因而合理地解决RFID系统中的数据碰撞问题成为了物联网研究过程中的重点和热点。标签数据碰撞实质是无线通信技术中的多路存取问题。在RFID系统中一般采用时分多址技术解决，即把信道容量按照时间先后顺序分配给多个用户，主要以基于ALOHA的概率型算法和基于树型结构的确定性算法为两大主要研究方向。基于树型结构的防碰撞算法能够准确识别出所有标签信息，避免了某些标签一直不能被读出的问题，其中又包括二进制搜索树算法和查询树算法两大类。基于树型结构的防碰撞算法整个查询过程可以用一棵多叉树表示，由阅读器不断发送查询命令和标签回复组成。阅读器发送的第一个命令对应于树根节点，接下来的查询以此为起点。本文针对标签数量众多时，整棵树查询将会较多的标签碰撞这一缺点，将一整棵大树分解为若干小的分支，在每一个小分支内查询标签，以减少阅读器的查询次数和标签碰撞发生的概率，从而提高标签数据众多情况下整个系统的性能，具体改进的内容如下：1．增加读取阅读范围内所有标签前缀阶段，每一个前缀代表整棵查询树的一个小分支。在具体读取标签信息之前，阅读器预先发送前缀查询命令以读取所有标签前缀，并把得到的标签前缀依次入栈保存；2．拥有相同前缀的标签同属于一个分支，从而把整棵查询树分为若干分支。阅读器对标签信息的读取以每一个前缀为起点，依次遍历每一分支，减少了标签数据碰撞发生的概率；3．在对每一个分支进行查询时，为了减少阅读器查询次数并降低时延，改进算法采用后退式搜索策略和动态搜索方式，以进一步提高整体系统性能。此外，对查询前缀长度的选择也做了讨论和分析，确定了系统性能最优时前缀长度的选择标准。最后，通过Matlab仿真平台对改进算法的有效性和优越性进行验证，主要考察阅读器查询次数、传输时延和系统吞吐率三方面的性能指标。仿真结果表明，改进算法在标签数量众多的环境下，能够有效减少查询次数，降低传输时延，提高系统吞吐率，性能良好，具有一定应用价值。更多还原

【Abstract】 As the product of information technology in the new times, the radio frequencyidentification (RFID) demonstrates great value of application. RFID is based on radiowave technology, and it makes use of space electromagnetic coupling to realize thewireless two-way data communication and identify some things automatically. Thedevelopment of RFID technology associates with growing-up of many other technologies,which has something to do with computer, wireless data communication, electromagnetism,cryptology and some other fields. Also it has lots of advantages, such as stronglypenetrability, convenient and efficient data reading, high degree of automation, easyoperation etc. For such unique advantages, RFID technology has been widely applied inindustrial manufacture and our daily life, so as to bring about lots of convenience for us.In recent years, the Internet of Things (IOT) has gained more and more recognition,so research on IOT becomes more and more deeply and widely. In the entire systemstructure of IOT, RFID is one of the key technologies of IOT’s sensing layer. For theapplication of IOT in the future, it will make full use of RFID to collect and transmitinformation. One advantage of RFID is multi-objective recognition, that means to identifylarge number of tags rapidly and accuracy among the reading field. But as the massive dataenvironment of IOT, when a lot of tags contend for the same channel to transmit their data,it will cause the problem of data collision, which have a significant influence on theintegrity and efficiency of data transmission. Then how to solve the problem of datacollision in RFID system reasonably becomes to the focus and hotspots of the research onIOT.Actually, the data collision is multi-access problem in wireless communication, andthe Time Division Multiple Access (TDMA) is generally adopted to solve this problem inRFID systems, it means distributing the channel capacity to lots of users according tochronological order. The main anti-collision algorithms are divided into two types: theprobabilistic algorithms based on ALOHA and the deterministic algorithms based on treestructure. The algorithms based on tree structure are able to identify all tags accuracy, thatavoid the problem of some tags cannot be identified. The deterministic algorithms includethe binary search algorithms and query tree algorithms.A multi-tree can be used to express the entire process of anti-collision algorithmsbased on tree structure. And it is composed of reader’s queries and tags’ responses. Thefirst query that is sent by the reader is the root node, which is the start of the query process.If the query process proceeds among the entire tree, it will cause lots of collision. In this paper, an improved anti-collision algorithm is promoted to conquer this problem. Theimproved algorithm divides the whole tree into some branches, among which to identifytags. As a result, the times of reader’s queries and tags collision will be reduced, and thenthe system efficiency is improved. The specific improved contents are as follows:1．A part is added to make sure of the prefixes that all tags have. Every prefix standsfor a branch of the query tree. Before reading information of tags, the reader sends acommand of prefix query, so all prefixes can be identified, and the prefixes will be pushedinto stack for saving.2．The tags that have the same prefix belong to the same branch, so the entire querytree is divided into some small branches. The identification process of tags starts fromevery prefix, then go through every branch. As a result, the probabilisty of tag collision isreduced.3．For reducing the times of reader’s queries and time delay, the back-track strategyand dynamic search are adopted when querying every branch. As a result, the performanceof the entire system will be improved.Furthermore, this paper also discuses and analyses how to select the prefix length.Then ascertain the standard of prefix length selection to achieve the optimal systemperformance.At last, we verify the effectiveness and superiority of the improved algorithm throughthe simulation platform of Matlab. The times of reader queries, the time delay and thethroughput are observed. As the result shows, when the number of tags is large, theimproved algorithm is able to effectively reduce the number of reader queries and timedelay, the throughput is increased at the same time. That means the improved algorithmhas a well performance, and it has value for application.更多还原