Enhancing Performance of IoT Sensor Network on Machine Learning Algorithms

Abstract

Internet of Things (IoT) technologies and machine learning algorithms are 
converging to give us new ways to view network performance and security 
in connected systems. With a growing trend of IoT sensor networks being 
deployed across a wide variety of industries , including smart cities and 
renewable energy infrastructure, its never been more vital to have robust, 
adaptive and intelligent management systems. In this article, we introduce 
the state of the art of using machine learning to optimize the operation of 
an IoT sensor network; in particular with respect to predictive maintenance, 
intrusion detection, and overall efficiency of the system. Since the birth of IoT 
sensor network, a long way has been traversed. First intended for relatively 
simple data collection and transmission, these networks now underpin 
enormous innovation across all sectors. The evolution of IoT sensor networks 
has been marked by several key developments: Today, sensor networks can 
reach down into the thousands, resulting in enormous data ecosystems. 
The inherent scale of this increased volume has proven to be a burden on 
data management and network optimization, and consequently requires 
more kinesic approaches to the sheer volume of information. Advances 
in sensor technology has lead to increasingly sophisticated, specialized 
and more diverse and precise data. The increasing availability of sensors 
covering a large range of applications—from environmental monitoring to 
instrumentation on industrial equipment—are generating an explosion 
of information, which can be used to improve decisions and the system  
performance.