【作者】 王晓东；

【导师】 杜新华；

【作者基本信息】 上海交通大学 ， 通信与信息系统， 2007， 硕士

【摘要】 射频识别技术(RFID,即Radio Frequency Identification)是从二十世纪八十年代开始走向成熟的一项自动识别技术。它利用射频方式进行非接触式的双向通信,交换数据从而达到识别的目的。由于大规模集成电路技术的成熟,射频识别系统的体积大大缩小,成本逐渐降低,从而进入了实用化阶段。在RFID领域,多个电子标签在识别时发生碰撞是射频识别技术中需要解决的关键问题。本文的研究重点就是改进防碰撞算法,提高标签的读取效率(单位时间内读取标签的个数)。目前,常用的解决该类问题的算法有:ALOHA算法、二进制搜索法、以及EPC Global的Class1 Gen2标准中所使用的Q-选择算法。本文通过引入动态实时处理的思想,对ALOHA算法进行了改进,通过数学计算,证明了改进后的算法在读取速率和正确读取概率上都比原算法有了较大的提高。对于二进制搜索法,本文通过引入动态和返回式的思想,证明了当读写器读取范围内的标签数量较多时,返回式的二进制算法的性能将有很大的提高,且标签数越多,这种优势也就越明显。对于Q-选择算法,本文提出将此算法中的C设定为随系统参数Q改变的值,其变化原则是C的取值范围在0.1至0.5间,并且C随着Q值的增大而减小。通过在NS-2中的建模仿真,将C可变与多种不可变的情况进行对比,证明了在C可变的情况下,系统的性能不会因为场景内标签数量的变化而出现明显的变化,平均的读取速率比C不可变时有了较大的提高。另外,本文还通过数学计算,证明了Q-选择算法比二进制搜索法的优越性,这种优越性尤其体现在当标签中的EPC信息量比较大的情况下。在原有的RFID系统中,同一时刻只能有一个读写器处于工作状态。本文提出了一种读写器的同步算法。通过将该算法在NS-2中建模仿真,并改变系统的各项参数,如负载大小,得到一系列仿真数据,证明了该算法的良好性能,极大的提高了系统读取标签的效率。更多还原

【Abstract】 RFID (Radio Frequency Identification) is an automatic identification technology which has been developed from 1980’s. It makes the identification through contactless duplex communication via microwave. Because of the maturity of Very Large Scale Integration, the volume and the cost of RFID device has been reduced, and it comes into practical phase.In RFID field, Tag collision is a key problem when many electric tags are identified. The key point of this paper is to improve the anti-collision algorithm and enhance the reading speed of RFID system. Now, the most frequently used algorithms are ALOHA, Binary-tree search, and Q-Selection which is specified in the EPC Global Class1 Generation2 standard. In this paper, we bring forward the dynamic and real-time process idea to improve the ALOHA algorithm. By mathematic calculation, we proved that the new dynamic and real-time algorithm has an improvement over the reading speed and capability. For the binary-tree search algorithm, we introduce the dynamic and back-track idea, and through calculation we proved that when there are many tags in the reading area the performance will be improved a lot. For the Q-Selection algorithm, we made the parameter C changeable when Q is changed, that means 0.1<C<0.5, and C is relatively large when Q is small and vice versa. Through simulation in NS-2, we proved that Q-Selection algorithm with making C changeable has an advantage over the one with C unchangeable. And we show that Q-Selection algorithm is better than binary-tree searching of backtracking in some scenarios where the bit length of EPC in the tag is long.In the past, there is only one reader can work at the same time. This paper presents a reader synchronization algorithm. Through modeling and simulation in NS-2 and changing different parameter such as the number of readers、system load and so on, we prove the good performance of this algorithm which will improve the efficiency a lot.更多还原