BICOT: Big Data Analysis Approach for Clustering Cloud based IoT Systems

Abstract

Internet of Things (IoT) envisions the connection of billions of devices over the Internet. The data produced by these huge amount of devices grow exponentially, so analyzing this big data with traditional methods is not viable. Recent cloud computing and virtualization technologies cope with these issues by processing and storing IoT data. Wireless sensor networks (WSNs) are big data sources of IoT systems which provides data collection from the environment. WSNs are used in various applications such as habitat monitoring, military surveillance and smart agriculture. Data transmission to the sink node is one of the essential requirements for WSNs. Clustering is a fundamental technique that is used for efficient data transmission, time synchronizaion, load balancing and security services. In this paper, we propose a clustering framework that we call BICOT for WSNs tailored for IoT systems. BICOT inputs large scale node position, transmission range and node energy data and outputs clustering information. Our first algorithm (BICOT-CDS) is based on connected dominating set (CDS) structure and aims to reduce the cluster count. Our second algorithm uses a weighted CDS (WCDS) approach that targets to select nodes with high energy as cluster heads. We implement these algorithms in ns2 simulator environment and measure cluster count and total weight of cluster head values. The algorithms are tested against node counts and average node degrees. From extensive simulation measurements, we obtain that the cluster count generated by BICOT-CDS is far more better than its counterparts and as the network size increases the proposed algorithm performs better. The cost of dominators produced by the BICOT-WCDS algorithm is significantly lower than its competitors. These findings show us that our proposed algorithms are favorable big data analysis approaches for cloud based IoT systems.

Clustering and parallel indexing of big IoT data in the fog‐cloud computing level

“…A variety of clustering methods have been proposed in literature. [47][48][49][50][51][52] Among the methods based on IoT data clustering, we cite the following methods. Lin et al 53 proposed a cloud storage approach to store Internet data that rely on the use of a lightweight data structure to visualize the data distribution and to calculate the size of each cluster.…”

“…A variety of clustering methods have been proposed in literature 47‐52 . Among the methods based on IoT data clustering, we cite the following methods.…”