Based On The Research And Application Of Column Stores Rfid Data Management Technology

RFID technology, also known as radio frequency identification technology, is a non-contact automatic identification technology. It's widely applied in logistics, security and other fields because RFID techonlogy can be operated simply in harsh environment without interference. However, compared with traditional data, RFID technology has many characteristics. For example, RFID data has simple tuple, loose structure, and RFID application creates large volume of redundant data in daily use. Due to these characteristics, RFID technology attracts much attention of insititue and industry in modeling, storage, query and other research fields.Considering the mass and redundancy of RFID data, this paper applied RFID technology to column-oriented storage model, which is able to deal effectively with redundant and loose-strutctured data. And we desinged an approach to process and query massive RFID data with MapReduce, a distributed data computing model. In addition, in order to compensate the deficiency that current column-oriented storage system can only support simple query, this paper designed and implemented a MapReduce-based column-oriented query framework, which can handle complex queries including equi-join. This paper also designed and implemented a MapReduce scheduling algorithm to improve MapReduce computing performence in dynamic scenario. Finally, we carried out experimental analysis and validation. Experiments approved that our MapReduce-based column-oriented query framework and improved MapRedcue scheduling algorithm can store and query massive and redundant RFID data efficiently in some scenario.