Mobile Information Systems

Volume 2019, Article ID 7542324, 15 pages

https://doi.org/10.1155/2019/7542324

## Target Detection Coverage Algorithm Based on 3D-Voronoi Partition for Three-Dimensional Wireless Sensor Networks

^{1}College of Computer Science and Engineering, Northwest Normal University, Lanzhou 730070, China^{2}Gansu Province Internet of Things Engineering Research Center, Lanzhou 730070, China

Correspondence should be addressed to Zhanjun Hao; moc.621@oahnujnahz

Received 17 October 2018; Accepted 14 February 2019; Published 10 March 2019

Academic Editor: Michele Garetto

Copyright © 2019 Xiaochao Dang et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

#### Abstract

The detection of target events is an important research area in the field of wireless sensor networks (WSNs). In recent years, many researchers have discussed the problem of WSN target coverage in a two-dimensional (2D) coordinate system. However, the target detection problem in a 3D coordinate system has not been investigated extensively, and it is difficult to improve the network coverage ratio while ensuring reliable performance of WSN. In addition, sensor nodes that are initially deployed randomly cannot achieve accurate target coverage in practice. Moreover, it is necessary to consider the energy consumption factor owing to the limited energy of the sensor node itself. Hence, with the objective of addressing the target event coverage problem of WSNs in 3D space applications, this paper proposes a target detection coverage algorithm based on 3D-Voronoi partitioning for WSNs (3D-VPCA) in order to ensure reliable performance of the entire network. First, we extend Voronoi division based on the 2D plane, which allows 3D-Voronoi partitioning of sensor nodes in 3D regions. Then, it is optimized according to the 3D-Voronoi neighbouring node partitioning characteristics and combined with the improved algorithm. Next, we set the priority coverage mechanism and introduce the correlation force between the target point and the sensor node in the algorithm, so that the sensor node can move to the target position for accurate coverage. Finally, we carry out related simulation experiments to evaluate the performance and accuracy of the proposed algorithm. The results show that the proposed algorithm can effectively improve the coverage performance of the network while ensuring a high overall coverage ratio.

#### 1. Introduction

A wireless sensor network (WSN) is a multihop ad hoc network consisting of a set of sensor nodes, small devices, etc., which are tasked with monitoring events in the region of interest and transmitting the collected data to the data centre for processing [1, 2]. The sensor nodes can be autonomous or stationary. Thus, we can organize the movement of the sensor nodes through self-deployment (e.g., by humans or mobile robots) [1]. In recent years, WSNs have witnessed many applications, such as target detection [3], target location [4], healthcare monitoring [5], and data collection [6]. Research on WSNs coverage is mainly classified into three branches: area coverage, barrier coverage, and target event coverage. Among them, target event coverage has long been an important research branch. We can deploy sensor nodes in the monitoring area to sense whether there are specific targets or events in practice. Hence, the main issues that we need to consider in the process of target coverage are as follows:(1)How to effectively improve the coverage and node utilization rate of the entire network(2)How to guarantee high coverage ratio and connectivity of the network, reduce the energy consumption of the nodes, and prolong the network life time(3)How to design algorithms or methods to ensure that the experimental environment is closer to a realistic 3D environment

In this study, we mainly investigate the target detection coverage of WSNs in 3D space. Previous studies on target coverage detection have achieved significant progress in terms of the theory, method, and experiment of using a 2D plane coordinate system to solve the 3D target coverage problem in a real-world environment, as discussed in the literature [7–9]. However, relatively few studies have investigated the coverage of WSNs directly applied to a 3D coordinate system. On the one hand, it is because the research difficulty increases with the dimension. On the other hand, the coverage problem of sensor nodes in a realistic 3D environment is often affected by the surrounding complex terrain and adverse weather conditions. In a real-world environment, sensor nodes are usually randomly deployed in the monitoring area, resulting in low utilization of the nodes. Hence, how to reduce the energy consumption of the sensor nodes and improve the coverage ratio of the network and the reliability of the algorithm are critical factors to be considered. In the literature [3], a more realistic signal detection model has been developed to obtain information on the target event and make the final decision by using a probabilistic decision model. Furthermore, a probabilistic detection algorithm [3] has been proposed to exploit the local measurements collected by the sensor nodes and improve the utilization rate of the nodes. In addition, a tracking framework based on Voronoi Tessellations has been proposed [10], whereby two mobility models are designed to control the coverage degree according to the existence of the target. Although this method can effectively monitor target events and discover redundant nodes, it is not suitable for realistic 3D conditions. A target-based Voronoi greedy algorithm (TV-Greedy) has been proposed [11] to find approximate optional solutions in order to improve the coverage quality of the network nodes while maximizing the effective energy consumption of the nodes. Although the TV-Greedy algorithm can effectively improve the robustness of the network, it is difficult to apply it to a 3D coordinate system. Thus, it is difficult to directly apply the traditional WSN coverage methods for a 2D coordinate system to a 3D coordinate system. The environment and position information of the target point in 3D space are more complicated. Hence, we need to redesign the deployment scheme of the sensor nodes in 3D space and propose relevant algorithms to achieve effective target coverage and a high coverage ratio of the network.

In summary, this study first designs and extends the traditional Voronoi diagram partitioning method and applies it to a 3D coordinate system, i.e., the 3D-Voronoi partitioning method is designed to solve the 3D space problem. Furthermore, the specific partitioning method is presented. Second, we propose an improved mobile algorithm to improve the detection effect and coverage ratio of the sensor network while addressing the omission problem of the target event to be monitored in the given space. In addition, we experimentally evaluate the network energy consumption as well as the accuracy and feasibility of the proposed algorithm in order to further reduce the network energy consumption. Moreover, we design a priority coverage mechanism and use the improved virtual force algorithm to achieve sensor node movement. Finally, we conduct experimental simulations and comparative analysis to further analyse the effectiveness of the algorithm. The main contributions of this article are as follows:(1)To the best of our knowledge, this is the first study to propose the application of the 3D-Voronoi partitioning method to target detection coverage of 3D WSNs.(2)We propose an improved 3D-Voronoi algorithm to ensure a high coverage ratio of the WSN on the basis of the energy consumption and connectivity factors of the network.(3)We optimize the traditional virtual force algorithm (VFA) to suit new environments and practical conditions. Furthermore, we conduct a full theoretical analysis of the algorithm and compare it with two other algorithms to verify its effectiveness and accuracy.

The remainder of this paper is organized as follows. Section 2 reviews the research progress and related studies on WSNs in recent years. Section 3 describes the design of the network model and the 3D-Voronoi partitioning model. In addition, it states the related definitions employed in this paper. Section 4 discusses how the related algorithms are designed and improved. Furthermore, it outlines the design steps of 3D-VPCA algorithm. Section 5 presents the theoretical analysis of the network coverage ratio and energy consumption of the proposed algorithm. Section 6 describes the simulations and experiments performed to compare the proposed algorithm with two other algorithms. Finally, Section 7 states the conclusions and briefly explores directions for future work.

#### 2. Related Works

In recent years, many studies and analyses have been carried out using the Voronoi plane to divide target events and sensor nodes. In particular, the optimal deployment of sensor networks is currently a research hotspot. Under some special circumstances, sensor nodes are often randomly distributed in a region. Optimizing the cost and utilization of limited resources has become the focus of current research. For example, in [12], distributed self-deployment schemes of mobile sensors have been proposed to solve the problem of efficiently deploying wireless sensor nodes. The authors designed two schemes by using a Voronoi diagram and a centroid, namely, Centroid (i.e., based on centroid scheme) and Dual-Centroid (i.e., based on dual-centroid scheme), to improve the speed and efficiency of node coverage holes. In [13], the authors proposed a deployment method for nodes in large-scale and high-density WSNs on the basis of centroid Voronoi Tessellation (CVT), which approximates the solution through the geometry of random points. Furthermore, the authors proposed a deployment plan based on the given characteristics of the study area in order to achieve near-ideal deployment. In some cases, owing to the initial random distribution of sensor nodes, there are coverage holes in the monitoring area. Hence, it is necessary to consider the survival time and cost of the network. Therefore, a two-phase coverage enhancing algorithm for hybrid WSNs was proposed in [14]. In the first phase, the authors used a differential evolution algorithm to compute the candidate’s target point positions in the mobile sensor nodes in order to improve the coverage ratio. In the second phase, they used an optimization scheme for the candidate’s target positions calculated in the first phase in order to reduce the moving distance of the mobile sensors. Finally, accurate filling of the coverage holes was achieved. Furthermore, the power supply module of the sensor node is a dry battery with limited energy, which can supply power only for a certain period of time. Hence, the energy consumption of nodes is a key research consideration. In [15], the authors discussed the energy consumption issue of target detection in WSNs and established a framework for evaluating performance indicators to analyse the performance of sensor networks as well as the quality of service. In [16], the author discussed the key issues of how to balance the target detection quality and lifetime of WSNs, and two target-monitoring schemes were proposed. One exploits the residual energy in the network and uses the adjustable sensing frequency in different regions to improve the monitoring quality. The other provides a method for calculating the optimal frequency value of the nodes with residual energy. Although the network energy consumption and monitoring quality factors have been considered in [15, 16], they have not been comprehensively considered or verified for a 3D coordinate system. Furthermore, the algorithms and methods proposed in the abovementioned studies are based on a 2D coordinate system, whereas actual scenarios are tested and verified in a 3D environment. Hence, some coverage methods for realistic 3D environments have also been proposed. For example, in [17], the authors proposed a distributed algorithm for mobile robotic sensors to allow self-deployment of sensor nodes in a 3D environment in order to achieve complete blanket coverage. This algorithm aims to achieve full coverage of the network while considering issues such as obstacles in the environment, and it minimizes the number of sensor nodes and mobile energy consumption. In [18], the authors pointed out that traditional probability coverage problems of WSNs mainly focus on 2D space. However, most practical applications of WSNs are in 3D space. Hence, in [18], the authors introduced a probability model of 3D WSNs and proposed a scheduling algorithm (PMCCA) that uses Voronoi division to control the scheduling of the probability model nodes in the target region. In [19], a 3D space deployment algorithm was proposed for continuous target tracking to overcome the dispersion problem of the traditional virtual force algorithm as well as the short-time tracking problem in the process of target tracking. In addition, the authors combined the internode force, the obstacle repulsive force, and the attractive force between the monitored-path and the tracking target into a virtual resultant force, so that the coverage of the target path is more continuous and sustained. Compared with the traditional virtual force algorithm, the abovementioned algorithm improves the effective time of the continuous target-tracking process and shortens the time of losing targets. In general, the algorithm or method proposed in [17–19] achieves effective detection of targets in 3D conditions. By contrast, our scheme is more direct and effective, and we have made more comprehensive considerations. In [20], the complete coverage problem of mobile sensor networks in a 3D environment was studied. The authors proposed a decentralized random algorithm to drive a group of mobile sensors on the vertices of a truncated octahedral grid for complete coverage of a bounded 3D area. However, the node utilization is not high, and this approach is not suitable for target event monitoring. In [21], a new network coverage and optimization control strategy based on a genetic algorithm was proposed to solve the deterministic coverage problem of sensor nodes. In addition, the author reduced the 3D coordinate system to a 2D plane in order to determine the fitness function of the relevant solution, and they used iteration to find the optimal solution. However, the abovementioned approach is based on the conditions of node deterministic deployment and is not applicable to the random deployment conditions considered in this paper.

In early coverage research on 2D-Voronoi partitioning, a distributed self-deployment protocol for mobile sensors was proposed [22] to calculate the correct location of the mobile node in order to solve the coverage vulnerability problem. Furthermore, the author considered the transition problem of nodes in the coverage area from dense to sparse and designed three sensor node mobility assistance algorithms. Experiments showed that these algorithms can achieve a higher coverage ratio in a short time with limited moving distance. Owing to the random deployment of a large number of sensor nodes in a real environment, the network degree of coverage is not high. In [23], a mobile sensor network coverage optimization algorithm based on virtual force perturbation and cuckoo search (VF-CS) was proposed. The Voronoi diagram was divided into sensor nodes to form their respective Thiessen polygons; then, a virtual force between the polygon vertices and the neighbour nodes was introduced as the perturbation factor of the node position update. Finally, the cuckoo algorithm was used to optimize the mobile coverage. In [24], to solve the problem of coverage holes in static WSNs, the authors proposed a triangle patch method to enhance the mobile node’s ability to repair coverage holes. Finally, the authors used the algorithm auxiliary information provided by the coverage edge nodes to move the nodes to the best candidate location. In [25], to reduce the cost of the K-coverage sleep-scheduling algorithm and ensure effective monitoring by the nodes, the prescheduling-based K-coverage group scheduling (PSKGS) and self-organized K-coverage scheduling (SKS) algorithm were proposed. Finally, the authors concluded that the PSKGS algorithm improves the monitoring quality and network lifetime through simulation experiments, while the SKS algorithm reduces the computational and communication costs of the nodes. Obviously, the research based on 2D-Voronoi algorithms has shown better results, but it can rarely be applied to a 3D coordinate system. Therefore, this paper extends traditional Voronoi studies [22–24] to 3D wireless sensor network target detection coverage.

In the subsequent experiments, we compare the proposed algorithm with the CSA algorithm and the RA algorithm. Although the three algorithms can achieve effective detection, the coverage ratio of our algorithm is shown to be optimal.

#### 3. Network Coverage and Voronoi Partitioning Method

##### 3.1. Network Coverage Model

This paper studies the problem of WSN target detection coverage in a 3D environment. Therefore, we assume that the sensing model of the sensor node is a spherical coverage area with node coordinate as the centre and the sensing range as the radius. Initially, it is assumed that sensor nodes are randomly scattered in a target area of size , and the set of nodes is . Owing to the random deployment of nodes in the initial stage, various problems, such as uneven distribution of sensor nodes, excessive energy consumption of nodes, and repeated coverage of some target points, may occur. Thus, the utilization of the node is reduced. Furthermore, some target points may be missing and not covered. As shown in Figure 1 below, some target points are in the coverage area of the nodes, and some are not covered by the nodes. Among them, the three black mesh spheres represent the coverage of three nodes, and the small black dots represent randomly distributed target point events. Furthermore, the node sensing model shown in Figure 2 represents the relationship between the communication radius of the node and its sensing radius. The proof in [26] shows that the connectivity between network nodes can be guaranteed when the communication radius of the node is at least two times the sensing radius, i.e., . Figure 2 shows the sensing model of the node, where the small balls represent as the sensing range of the node , and the large balls represent the communication range with as the node .