Applying a neural network to recover missed RFID readings

Authors:
Peter Darcy;Bela Stantic;Abdul Sattar
Affiliations:
Griffith University, Queensland, Australia;Griffith University, Queensland, Australia;Griffith University, Queensland, Australia
Venue:
ACSC '10 Proceedings of the Thirty-Third Australasian Conferenc on Computer Science - Volume 102
Year:
2010

Citing 7
Cited 1

Neural Network Training Using Genetic Algorithms

Neural Network Training Using Genetic Algorithms
Temporal management of RFID data

VLDB '05 Proceedings of the 31st international conference on Very large data bases
Adaptive cleaning for RFID data streams

VLDB '06 Proceedings of the 32nd international conference on Very large data bases
A deferred cleansing method for RFID data analytics

VLDB '06 Proceedings of the 32nd international conference on Very large data bases
An investigation of the modified direction feature for cursive character recognition

Pattern Recognition
Managing RFID data

VLDB '04 Proceedings of the Thirtieth international conference on Very large data bases - Volume 30
Short survey: Taxonomy and survey of RFID anti-collision protocols

Computer Communications

An intelligent approach to handle False-Positive Radio Frequency Identification Anomalies

Intelligent Data Analysis

Quantified Score

Hi-index	0.00

Visualization

Abstract

Since the emergence of Radio Frequency Identification technology (RFID), the community has been promised a cost effective and efficient means of identifying and tracking large sums of items with relative ease. Unfortunately, due to the unreliable nature of the passive architecture, the RFID revolution has been reduced to a fraction of its intended audience due to anomalies such as missed readings. Previous work within this field of study have focused on restoring the data at the recording phase which we believe does not allow enough evidence for consecutive missed readings to be corrected. In this study, we propose a methodology of intelligently imputing missing observations through the use of an Artificial Neural Network (ANN) in a static environment. Through experimentation, we discover the most effective algorithm to train the network is via genetic training with a high chromosome population. We also establish that the ANN restores a cleaner data set than other intelligent classifier methodologies in the majority of the test cases especially when faced with large amounts of missing data.