Desktop version

Home arrow Computer Science

  • Increase font
  • Decrease font

<<   CONTENTS   >>

An Energy-Efficient Quasi-Oppositional Krill Herd Algorithm-Based Clustering Protocol for Internet of Things Sensor Networks


Generally, Internet of Things (IoT) models are deployed actively and offer massive facilities for humans and business applications. By default, the IoT devices would be interlinked and communicates with one another by interchanging the accumulated sensor Information. Hence, the performance metrics like operating systems (OS), interacting protocols, and hardware units are applicable to provide enormous promising issues in developing IoT networks. IoT middleware environments and studies have been developed for resolving these issues [1]. For middleware platforms, the combined developing platform has been applied, and the response time from devices is limited by resource management objectives. Such middleware environments are mainly applied for minimizing the energy utilization, as the machines in IoT is often implemented on battery power. Hence, sensing tasks utilizes sensing period strategies. In absence of data variation, the devices are computed on the basis of predefined sensing period that consumes maximum energy by transmitting unwanted data resources.

The modern tools involved in IoT systems like sensors, actuators, power modules, tools, interaction methods, etc. The basic architecture of IoT sensor networks is shown in Figure 11.1. It is operated with the help of power that exists in incorporated battery. The operational features of devices are composed of massive effect on power application value. On the contrary, a wireless sensor network (WSN) has an exclusive base station (BS), where the devices are composed of an IoT network that interacts with one another with no limitation; hence, the drastic difference of IoT services could be offered to the users. In order to accomplish this, the devices should be capable of transmitting or interchanging the data with related devices. As the sensors, transmitter and actuator devices have been applied in this application, and then the energy utilization becomes more sensible and higher. Releasing maximum energy in devices is composed of a negative impact in prolonged IoT networks, and subsequently the earlier studies are concentrated on the effective way of utilizing energy, and examining the low power for sensing task, minimal energy communications, and extended battery application with no replacement of batteries or recharging the batteries [2].

When the energy is released completely, it results in expiry of a node and minimizes the network duration. In order to resolve these limitations, studies developed on storing node energy and extending the network duration that are considered to be the major issues in WSN [3]. In addition, clustering model could minimize the power application of nodes and elongate the networks existence [4, 5]. A WSN network is classified as various subsets and

Architecture of IoT-sensor networks

FIGURE 11.1 Architecture of IoT-sensor networks.

so called as clusters. The nodes present in identical cluster are adjacent to one another. For all clusters, a single node is named as cluster head (CH), and alternate nodes are termed as the cluster members (CMs). In spite of interacting with sink in a direct manner, members are applied for sending the sensed information to CHs where the collected data is transmitted to sink node. When compared to earlier planar routing models, LEACH is applicable to offer maximized network lifetime [6,7]. Therefore, LEACH decides the CHs on the basis of the value produced by a node and previous fixed threshold that develops massive clusters and CH election is carried out in a random manner. But LEACH cannot ensure the optimal clustering model from energy management of the whole network.

Using the development of a LEACH, massive WSN clustering technologies are presented. It is composed of various clustering models on the basis of various clustering metrics like network architecture, communication method, topology, and stable routing. Here, the clustering methods are divided into two classes: independent and cooperative, on the basis of BS that is involved in clustering process. Autonomous clustering models on network nodes with no sink s contribution, and control data has been interchanged over all the nodes. The predefined clustering models can extend the network lifetime; however, it suffers from limitations with respect to working function. The independent methods are implemented on all nodes, and massive control data should be interchanged which results in extra power application. Though the cooperative models are involved in limiting the power application of control data exchange, this fails to obtain optimal clustering from complete network as it is composed of several parameters that affect the clustering and such attributes conflict mutually. Specifically, using the deployment of nature-based methods, the clustering techniques have been applied for enhancing clustering function recently [8].

An effectual node ranking LEACH (NRLEACH) for WSN has been projected. Here, the nodes are classified as two phases like clustering setup and data transmission. The NR model has been applied for CH election. NR approach has also been applied for calculating the grade of all nodes by count of iterations, which depends upon different parameters. Energy-centered clustering-based routing technique is presented in Ref. [9], which classifies the complete network to various static grids, and the nodes present in all grids are comprised of a cluster. The node estimates the rank based on the residual energy (RE) and maximum distance to alternate nodes in a cluster. Initially, each node telecasts the rank value for other nodes similarly. The node with maximum rank value can be selected as a CH. A new fuzzy clustering model is developed in Ref. [10] for 3D WSN named as FCM-3 WSN. Here, the developers have introduced an arithmetic approach of clustering in 3D WSN that considers the network power application, communication metrics, and the features of 3D network. The Lagrange multiplier (LM) approach has also been applied for determining the cluster centres as well as the membership matrix used for clustering operation.

A novel energy-efficient centroid-based routing technique (EECRP) is presented for WSN. In EECRP, the application of “energy centroid” has been presented. In Ref. [11], Zhang et al. have provided an enhanced LEACH method for WSN for optimizing the CH selection. First, the BS telecasts the initialization message to the all systems. Once the initialization message is received, nodes determine the distance from corresponding BS on the basis of obtained signal strength and save the data of adjacent nodes in a table. In Ref. [8], a novel power-effective multihop routing method (MR) has been developed. In MR, nodes are classified into three classes: CHs, member nodes, and independent nodes (IN).

A new particle swarm optimization (PSO) relied energy-efficient CH selection model for WSN (PSO-ECHS) has been presented. In PSO-ECHS, the applied attributes assumed, like intracluster distance, distance from CHs to BS, and the impact of RE on clustering outcome, and a linear integration of such variables have been applied in fitness function (FF) of particles.

A PSO-based method for routing as well as clustering in WSN has been presented in Ref. [12]. In case of clustering technique, the power application of CHs and CM nodes are considered. While using PSO for clustering the WSN, the coding model of particles is the dimension of particles and similar to count of network nodes, and location component of all dimension of particles is meant to be arbitrary value r that follows uniform distribution from (0,1). An integer к is attained from r using transformation formula, and к is declared as CH-ID where the nodes of particle position. Moreover, an FF is mainly applied for reducing the power application of data transmission and manages the RE in transmission path.

In Ref. [13], Kuila and Jana presented a power-efficient clustering as well as routing methodologies for WSN such as PSO model. Here, the particle encoding approach for clustering is similar, but the objective of FF is applied to build a CH with maximized network lifetime and CM has smaller average distance to CH. In Ref. [14], Wang et al. recommended an enhanced routing framework with certain clustering with the help of PSO approach for heterogeneous WSN (EC-PSO). Initially, CHs are selected on the basis of geographical position. If the network is implemented for a period and power of nodes is not the same, EC-PSO is utilized for clustering the WSN.

This chapter presents an efficient metaheuristic quasi-oppositional krill herd (QOKHC) based clustering algorithm for IoT sensor networks. The proposed QOKHC algorithm involves the selection of CHs and organizes clusters. The presented model integrates the quasi-oppositional based learning (QOBL) in the krill herd (KH) algorithm to increase the convergence rate. The efficiency of the QOKHC-based clustering technique has been assessed and the results are examined under diverse measures: RE, network lifetime, alive node analysis, and the number of packets transmitted to BS.

<<   CONTENTS   >>

Related topics