An energy-efficient artificial bee colony-based clustering in the internet of things

Highlights

• Improving the tradeoff between energy consumption and transmission delay in IoT.

• Exploiting artificial bee colony to select the efficient cluster-heads in IoT.

• Considering energy, neighbors and distance as the criteria for cluster-head selection.

• Providing an artificial bee colony-based mechanism for clustering devices efficiently.

• Considering distance and data volume as the criteria for clustering IoT devices.

Abstract

Wireless communication on the Internet of Things (IoT) requires context-aware data transmission protocols. Developing an energy-efficient clustering mechanism is the primary challenge in data transmission over IoT. The existing approaches struggle with the short lifetime of IoT, imbalance load distribution, and high transmission delay. This paper proposes a novel cluster-head selection and clustering mechanism on IoT. It is composed of two main phases. The first phase selects the near-optimal cluster-heads using Artificial Bee Colony (ABC) algorithm. Performance criteria include the residual energy of the devices, the number of neighbors, Euclidean distance between devices and the sink, and Euclidean distance between each device and its neighbors. The principal objective of the second phase is to group devices into some clusters based on Euclidean distance between each cluster-head and its members, and the data volume generated by clusters. Simulation results verify that our mechanism improves energy consumption, lifetime, and transmission delay.

Introduction

The interconnections of smart devices on the Internet of things (IoT) provides an intelligent landscape for a dynamic world. IoT refers to the interconnection of the networks between heterogeneous wireless sensors and routine physical devices scattered in a monitoring area to sense, gather, share and transmit data throughout the system, and provide a smarter life for human beings [1]. Meanwhile, from the recent advances in technologies and protocols, IoT devices are equipped with sensing, processing, communication, interaction, and corporation capabilities. According to these properties, IoT has penetrated pervasively into the wide-scale aspects of human life, including smart healthcare systems and personal monitoring, smart cities (homes, buildings, traffic monitoring, and urban computing), intelligent commercial, and smart industrial systems [2].

Many IoT devices are powered by batteries with a limited lifetime and deployed in remote areas. Therefore, one of the main concerns while using IoT is energy limitation, as communication and computations on devices might quickly exhaust their battery resources [3]. Thus, providing energy-efficient data processing, aggregation, and transmission mechanisms play a vital role in IoT applications. The amount of energy consumed for wireless communications in the internet-based systems is much more than the processing costs [4]. Therefore, one of the significant challenges on IoT is to achieve an energy-efficient data transmission mechanism from the source to the destination devices. Clustering is the primary operation required for appropriate data transmission on IoT. It is defined as a creative process that groups devices in some clusters and determines a cluster-head for each of them to improve resource consumption [5].

There are different heuristic [6], metaheuristic [7], and fuzzy-based [8] clustering mechanisms in the literature. The heuristic mechanisms try to provide an approximate solution to a specific problem. However, the clustering challenge requires approximation problem-independent solutions for which it can be proved how close the final solution is to the optimal one. Also, the primary purpose of most heuristic mechanisms is to reduce the number of clusters, which increase injected traffic into the cluster-heads and reduce IoT lifetime. The metaheuristic clustering mechanisms focus on the distance and residual energy of devices as the performance criteria, while they ignore the number of one-hop neighbors and data volume. Besides, some metaheuristic mechanisms have suggested the cluster-head selection scheme, while they offer no idea about the device allocation to the clusters. Finally, the solutions from fuzzy-based mechanisms are perceived based on assumptions, while setting exact assumptions, rules, and functions is a complicated process. Furthermore, validation and verification of the fuzzy-based systems need extensive tests. So, it is more efficient to provide a metaheuristic cluster-head selection and clustering mechanism on IoT by considering the data volume generated by each cluster, residual energy of devices, the number of one-hop neighbors, the distance between devices and the sink, and the distance between each device and its one-hop neighbors as the performance criteria.

Since clustering on IoT is an NP-hard problem, metaheuristic algorithms are a good idea to tackle this challenge. Artificial Bee Colony (ABC) is one of the well-known metaheuristic algorithms, which does not need the parameter-setting to address the issues. Also, it yields better performance in comparison to others [9]. So, this paper proposes a novel mechanism for cluster-head selection and device clustering on IoT, which is named ABC-based Device Clustering (ABC-DC). The presented mechanism consists of two phases. The first phase aims to select the near-optimal cluster-heads using the ABC. It considers the residual energy of devices, the number of one-hop neighbors, Euclidean distance between devices and sink, and Euclidean distance between each device and its one-hop neighbors as the performance criteria.

The principal objective of the second phase is to group devices into some clusters. It also exploits the advantages of the ABC to enhance tradeoff between the performance criteria, including energy consumption of the whole system, and data transmission delay. The Euclidean distance between each cluster-head and its members and the data volume generated by each cluster are employed as the performance criteria. Simulation results verify that the proposed mechanism improves IoT lifetime by reducing the energy consumption of devices and data transmission delay in comparison with the recent state-of-the-art approaches.

The fundamental contributions are summarized as follows:

• Developing an efficient metaheuristic clustering mechanism on IoT to tradeoff between performance factors, including energy consumption and data transmission delay.

• Developing a flexible clustering mechanism to comply with the requirements of the different IoT applications.

• Exploiting the advantages of ABC to select the near-optimal cluster-heads considering the residual energy of the devices, the number of one-hop neighbors, Euclidean distance between the devices and sink, and Euclidean distance between each device and its one-hop neighbors as the performance criteria.

• Developing an ABC-based mechanism to group IoT devices in some near-optimal clusters considering the Euclidean distance between each cluster-head and its members, and the data volume generated by each cluster as the performance criteria.

The rest of this paper is organized as follows: Section 2 explains the previous clustering mechanisms on IoT. Then, some practical criteria about the IoT modeling, and energy depletion pattern are described in Section 3. After that, the proposed mechanism is discussed in Section 4. Section 5 evaluates the performance of the ABC-DC and its features in comparison with recent state-of-the-art approaches. Finally, Section 6 represents the conclusion part.