An Overview of Deep Learning Techniques for Big Data IoT Applications

Abstract

ABSTRACT We are in the era of internet of things (IoT) technology where billions of smart devices communicate over the internet. The rapid development of the IoT has led to generation of huge amounts of data across broad range of fields. Analyzing this big IoT data is crucial for extracting valuable insights. These insights enable predictive vision, which in turn guide control decisions to enhance system performance and improve living standards. Although many deep learning (DL) models have been developed for IoT data analysis, existing literature often underlines domain‐specific applications, overseeing the need for a universal model selection framework. This paper aims to bridge that gap by offering a concise overview and a comparative framework to guide the selection of DL models based on IoT data characteristics, application needs, and hardware limitations. This paper presents an overview of IoT architecture and characteristics of IoT big data. It also discusses the computing infrastructures used for IoT data analytics. The role of deep learning in IoT big data analytics is then outlined, followed by a discussion of key deep learning model categories to process IoT data. Various IoT applications employing DL methods are presented. Finally, the challenges faced during the development of smart IoT systems, along with the future research direction, are outlined.