An AM-MCMC cleaning algorithm for multi-reader RFID redundant data

Yinju Lu, Guoquan Shan

COMPUTER MODELLING & NEW TECHNOLOGIES 2014 18(12A) 260-269

College of Information and Engineering, Zhongzhou University, Zhengzhou, China

The inherent characteristics of RFID device and the environmental noise cause the uncertainty of RFID raw data and, in the RFID event detection, decrease accuracy of the query results. In this paper, the recognition model of RFID reader is defined and, by using the maximum entropy method, 3-state recognition model in this recognition model has been proved to have the optimal performance. Using Bayesian principle, the posterior probability distribution of parameters to be estimated can be got from the condition likelihood observed and prior distribution of unknown parameters. Based on adaptive sampler, a Markov Chain Monte Carlo (MCMC) simulation is proposed to do data cleaning on the redundant data from RFID multi-reader. The simulation test results, carried on a large number of simulation data, verify the accuracy and efficiency of the proposed data cleaning algorithm.