分布式RFID系统的数据处理技术研究

发布时间：2019-01-09 06:54

【摘要】：RFID技术在近年取得了快速的发展,在多行业获得了广泛应用。近些年出现的分布式RFID为RFID技术开拓了新的技术空间。RFID的迅速发展和应用场景的多样化给RFID数据处理技术提出了更高的要求。如何快速有效的对RFID的数据进行清理以及分类是RFID技术进一步发展的重要环节。为解决分布式RFID数据处理的问题,论文针对分布式RFID的数据清理以及分类问题,完成以下工作:针对分布式RFID数据结构的新特性设计了数据处理系统,对数据进行集中处理,包括数据的去重和归类。在天线数目多于十个时,系统整体性能相较于采用传统处理系统有明显的提升。系统借鉴MVC结构进行设计,各模块之间耦合程度较低,便于功能的开发和扩展,增加了系统的稳定性。针对RFID数据去重问题,论文对传统的SNM算法进行改进。通过分析SNM算法的影响因素,对实验中理想状况和读取效果不佳的状况分别进行讨论,得到两种情况下最佳窗口取值大小的区域。在对应区域内,算法的处理性能有10%到15%的提升。论文采用匹配树算法对匹配过程进行改进,在解决RFID数据归类问题上相较于自动机算法,性能提高了35%左右。进一步引入剪枝算法对匹配树进行剪枝,在满足一定的概率阈值的情况下,对算法进行了进一步的优化。论文对各模块的功能进行了性能测试,用以验证算法以及模块的效果。通过性能测试,证明改进的算法在性能上相较于改进之前性能有明显的提升,充分的证明RFID数据处理的研究价值。在论文的最后,对RFID数据处理技术进行了总结和展望,展望结合机器学习对系统进行进一步改进的可能性,并且明确了针对不同RFID的应用场景开发更加普适的数据处理系统,从而实现完备的数据存储的系统。
[Abstract]:RFID technology has made rapid development in recent years, and has been widely used in many industries. The emergence of distributed RFID in recent years has opened up a new technological space for RFID technology. The rapid development of RFID and the diversification of application scenarios put forward higher requirements for RFID data processing technology. How to quickly and effectively clean up and classify the RFID data is an important step in the further development of RFID technology. In order to solve the problem of distributed RFID data processing, this paper aims at data cleaning and classification of distributed RFID, accomplishes the following work: according to the new characteristics of distributed RFID data structure, a data processing system is designed, and the data is processed centrally. Including data removal and classification. When the number of antennas is more than 10, the overall performance of the system is significantly improved compared with the traditional processing system. The system is designed by using MVC structure for reference. The coupling degree of each module is low, which is convenient for the development and expansion of functions, and increases the stability of the system. Aiming at the problem of RFID data de-duplication, this paper improves the traditional SNM algorithm. By analyzing the influence factors of the SNM algorithm, the ideal condition and the bad reading effect in the experiment are discussed respectively, and the region of the optimal window size is obtained under the two kinds of conditions. In the corresponding region, the processing performance of the algorithm is improved by 10% to 15%. In this paper, the matching tree algorithm is used to improve the matching process. Compared with the automata algorithm, the performance of the algorithm is improved by about 35% in solving the problem of RFID data classification. Furthermore, the pruning algorithm is introduced to prune the matching tree, and the algorithm is further optimized when a certain probability threshold is satisfied. The function of each module is tested to verify the algorithm and the effect of the module. The performance tests show that the performance of the improved algorithm is significantly improved compared with that before the improvement, and the research value of RFID data processing is fully proved. At the end of the paper, the RFID data processing technology is summarized and prospected, the possibility of further improving the system with machine learning is prospected, and the more suitable data processing system is developed for different RFID application scenarios. In order to achieve a complete data storage system.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP391.44

【参考文献】