Big Data Mining Techniques In IoT, Challenges and Architectures

Abstract

Nowadays, data is globally viewed as the most valuable resource, and the Internet of Things (IoT) has been playing an essential role ever since the time it emerged. Modern data sets are so complicated that cannot be handled by traditional software and hardware. Three major characteristics of the present time generated data are volume, velocity and variety, which have resulted in developing a concept called big data. Such characteristics have turned the routines of receiving, storing, processing, analyzing, and visualizing big data into a challenging issue. In the current competitive world, analyzing big data is critically important. The significance of big data doesn’t refer to the amount of data which is accessed by a company or organization, rather it depends on how the question data is used. Processing and analyzing the collected data helps enterprises to gain the due insights and benefit from them compatible with strategic decisions. Over the past few years, some novel frameworks and tools have been presented for storing, processing and analyzing big data so that their relevant know-how and thus, working with such large-scale data can provide the specialists in this field with various research areas and job opportunities. This paper has addressed big data in the IoT in which the issues about data mining architectures have been discussed. One of the prominent architectures raised in this field is the IoT-based multi-layered data mining model which is divided into four layers: data collection, data management, event management, and data processing service. Another architecture considered in this paper is the distributed data mining model whose main goal is pre-processing the distributed data before being submitted to the central receiver (core infrastructure) in order to reduce energy consumption in the central nodes. Grid-based data mining infrastructure in the IoT pursuing the objective to focus on the strategies to increase portability and situational awareness is another model which has been dealt with in the references. Another architecture is the data mining model that predicts the integration of several technologies in the IoT. In this architecture, the integration of several technologies including cloud computing, grid computing, pervasive networks, secure networks, etc. are dealt with. Another architecture under discussion is the IoT-based data mining on the cloud computing platform whose objective is to integrate the layers of extraction, management, exploration, and interpretation. Finally, a three-dimensional five-layered architecture was proposed for mining big data in the IoT, the main idea of which is a three-dimensional device-layer architecture consisting of the device layer, raw data layer, data collection layer, data processing layer, and service layer. The other two dimensions cover the issues like complying with the standards, security and privacy, data management, data perception, and data interpretation.