Abstract:
The recent expansion in the field of Internet of Things (IoT) and Big Data is providing a large
production potential in the course of the new era of smart urban. The IoT infrastructure for the
design of smart urban consists of devices, objects, sensors and citizens producing huge gigantic
data (i.e. Big Data). IoT-based smart urban environment provides the digital traces of human and
objects that can be analyzed. The major aim of smart urban is to efficiently utilize the data to
administer and resolve the issuesconfront by recent smart cities regarding data processing.This
study presents Hadoop-based Big Data analytics architecture to address the challenges in data
generated in IoT environment. The proposed architecture is based on customization of Hadoop
architecture and external entities to efficiently process Big Data generated in IoT environment.
The proposed scheme is comprised of Big Data loading into Hadoop and Big Data processing.
The existing solutions provide manual and serial data injection into Hadoop. Moreover, the
existing solutions do not tackle the communication overhead efficiently that affects the
processing. Data loading and storing is performed by proposing parallel and utility-oriented
solution based on multiple attributes. Unlike traditional MapReduce architecture, customized
YARN-based cluster management solution is provided to manage the cluster resourcesefficiently
and process the data using Map-Reduce algorithm separately. The proposed architecture is tested
with a variety of reliable datasets using Hadoop framework. The comparison of proposed
architecture with existing solutions and default architecture of Hadoop is provided to verify and
reveal that the proposed architecture is more efficient then existing smart urban architecture
using Big Data analytics for processing data produced in IoT-environment.
