DEVELOPMENT OF AN ENHANCED DATA ANNOTATION AND FILTERINGTECHNIQUE FOR IOT SENSORS
No Thumbnail Available
Date
2019-12-12
Journal Title
Journal ISSN
Volume Title
Publisher
Ahmadu Bello University Zaria
Abstract
Internet of Things (IoT) applications depend on data that are meaningful to the machine to
efficiently function. Amongst other sources, data are generated by different types of sensors such
as proximity sensor, pressure sensor, temperature sensor and ultrasonic sensor. The diversity of
these sensors reflects on the data they generate. As a result, IoT applications encounter
challenges understanding and processing these data. The most recent solution to this problem is
data filtering and annotation on gateways. This has also resulted into a bottleneck processing
thereby causing delay and inconsistencies in processed data. Consequently, an enhanced data
preparation and annotation technique is proposed. This approach uses a distributed programming
model for sensory data processing. The proposed approach seeks to develop a Hadoop
MapReduce algorithm which efficiently filters and annotates sensory data in a distributed
manner. To evaluate the feasibility of the proposed approach, data generated by sensors are
stored on Hadoop Distributed File System (HDFS) and are processed by a MapReduce job.
Semantic Web technologies such as Extensible Markup Language (XML) and Resource
Description Framework (RDF) were used for the data annotation. Two categories of experiments
were conducted and comparison between the proposed system and the existing system were done
based on data size and processing time. This dissertation concludes that the proposed system has
61.65% processing time and 13.41% data size enhancement respectively over the existing system