Sage Journals: Discover world-class research

Abstract

Wireless sensor networks are a cornerstone of the Internet of things with many applications. An important aspect of such applications is target tracking using self-positioned known sensor nodes. Over the years, many schemes have been proposed to locate and track the target path. However, accuracy and reliable tracking remain an open area of research. In this article, we propose a dynamic cooperative multilateral sensing scheme for indoor industrial environments to improve target localization and tracking accuracy. The scheme is designed to select reliable nodes based on the distance between nodes within-cluster and to the target for reduced positioning error. Furthermore, a cluster node is dynamically selected based on distance from the base station. We simulate the proposed technique in scenarios with tracking at regular intervals and with the complete path. Furthermore, the performance of the scheme is also tested under different sensor coverage areas. The results show that the proposed scheme provides better target tracking with up to 19% higher accuracy in comparison to the traditional trilateration scheme.

Keywords

Wireless sensor networks industrial Internet of things target tracking cooperative sensing

Introduction

With rapid advancements in technology, sensor nodes (SNs) are becoming smaller while consuming less energy. They are capable of connecting to one another in an ad hoc manner forming a single unit referred to as wireless sensor network (WSN).¹ Often the nodes are dispersed either randomly or uniformly over an area under observation for the purpose of monitoring and analysis. Notably, the network is cost-effective with minimal human intervention for many environmental and habitat monitoring applications, for instance, animal and bird habitat tracking,^2,3 agriculture and industrial land monitoring,^4,5 smart irrigation,^6,7 flood control and monitoring,⁸ underwater and space exploration exploration,^9,10 environmental pollution,¹¹ and health monitoring.^12,13

In many of these aforementioned applications, information coming from within the WSN is only useful when the referred location is known, for instance, during search and rescue, target localization, monitoring, and so on. More specifically, target tracking in WSN estimates the location and direction of the target using the information received by SNs. That is, the SNs gather the target signatures and forward it to the cluster head (CH).^14,15 Once the information is received at the CH, it is processed to predict target location and forwarded to the base station (BS). In all, the steps involved are target detection and tracking, tracking information forwarding, and target location prediction. The resulting system providing an accurate location of the target (indoor and outdoor) opens up new applications for crisis and commercial services, for example, rescue operation,¹⁶ forest fire monitoring,¹⁷ location known vehicle,¹⁸ shipment tracking,¹⁹ navigation,^20,21 health monitoring,²² home automation,^23,24 and so on. In the end, making daily life easier in terms of safety and efficiency.

One important concept emerging from these advancements is the industrial Internet of things (IIoT) that refer to a network of connected devices, sensors, and actuators forwarding data in the network without human interaction.^25,26 This real-time data forwarding for analysis and decision making is considered the main facet of efficient industrial automation. But it all starts with target tracking, a vital step in any IIoT application. Over the years, researchers have proposed multiple tracking frameworks for efficient scheduling of industrial equipment, effective data gathering, real-time data processing, target localization, and so on (Table 1).

Table 1.

Comparison of different techniques.

Authors	Target sampling rate	Target velocity	Target trajectory	Clustering technique	Energy optimized	Accuracy obtained	Network optimized	Simulator	Sensor deployment
Khan et al.²⁷	NA	NA	M	Static	Yes	Yes	NA	NS2	Random
Zhang et al.²⁸	50 s	NA	RW	Dynamic	Yes	Yes	NA	NA	Random
Khakpour et al.²⁹	600 s	25–70 m/s	H	Dynamic	Yes	Yes	NA	NA	On vehicle
Darabkh et al.³⁰	NA	0–17 m/s	RW	Dynamic	Yes	Yes	Yes	MATLAB	Uniform
Rouhani et al.³¹	NA	5–10 m/s	RW	Static	Yes	NA	NA	Castalia	Random
Luo et al.³²	60 s	1.5 m/s	RW	Static	NA	Yes	NA	MATLAB	Random
Bhowmik and Giri³³	NA	2–20 m/s	RW	None	Yes	NA	NA	J-Sim and MATLAB	Uniform
Abid et al.³⁴	NA	NA	R	None	NA	Yes	NA	NA	Star topology
Yu³⁵	100 s	NA	R	None	NA	Yes	NA	NA	Random
Jiang et al.³⁶	25 s	0.127–0.28 m/s	RW	None	NA	Yes	NA	COTS RFID	RFID
Xiao et al.³⁷	NA	NA	RW	None	NA	Yes	NA	OTS RFID	RFID
Chiani et al.³⁸	61 s	1.5 m/s	RW	Hybrid	NA	Yes	NA	NA	Random
Lu et al.³⁹	100 s	3 m/s	R	NA	Yes	Yes	NA	NS3	Random
Proposed	50–200 s	NA	RW/M/S	Dynamic	NA	Yes	Yes	MATLAB	Uniform

H: highway path; M: Manhattan; R: random; RFID: radio-frequency identification; RW: random waypoint; S: spiral.

Problem statement

Evidently, target tracking is inaccurate when done using a fewer number of SNs. But it becomes even more challenging when tracked frequently for maximum accuracy with scarce resources. Traditional schemes are centralized, compute-intensive, and complex; thus, inapplicable to WSN scenario due to limited processing capability and power constraint of SNs. The motivation of this study is to develop a fast and precise tracking algorithm that is self-sustained in terms of communication and processing. The main contributions of this article are as follows:

Propose a cooperative approach integrating cross-grid and region-overlapping scheme for indoor industrial environments with the goal to improve target localization and tracking accuracy.

Design a technique to select reliable nodes based on the distance between nodes within-cluster and to the target for reduced positioning error. Furthermore, a cluster node is dynamically selected based on distance from the BS.

Evaluate the proposed approach in terms of positioning and tracking error. Three most commonly used mobility models are used to benchmark the proposed work. Furthermore, the proposed work is compared with the conventional trilateration technique.

The rest of this article is organized as follows. Related work is presented in section “Related work.” System model is formulated in section “System model.” Section “Design and implementation” details the proposed algorithm design and implementation. The framework evaluation is covered in section “Performance evaluation” followed by discussion in section “Discussion.” Finally, the conclusion is presented in section “Conclusion.”

Related work

This section covers the existing contribution for target tracking. In this section, we have categorized the existing techniques in terms of clustering and non-clustering. However, a work on IIoT is added under separate headings.

Clustering-based target tracking

Over the years, researchers have proposed multiple techniques to improve the efficiency of WSNs. Among these techniques network clustering is widely used. It involves the initiation of cluster topology to divide the network into clusters and responsibilities involving the definition of nodes per cluster, coverage area, CH selection, and so on. The techniques are further categorized into static and dynamic clustering. In static clustering, the whole network is divided into clusters as soon as network become active, for instance, a trilateration-based static clustering scheme proposed in Khan et al.²⁷ Here, each SN calculates its distance to the target using a received signal strength indicator (RSSI) and forwards it to the CH. Using this information, the CH predicts the target location. The mechanism fails to cater to target path changes with no recovery mechanism. To overcome this, in Rouhani and Haghighat,³¹ a prediction-based target tracking scheme comprising three steps, clustering, tracking, and prediction, are proposed. In the first step, the network is divided into static clusters with CHs elected based on residual energy or distance to the BS. Next, a trilateration-based mechanism is used to track the target followed by a linear prediction (LP) model to predict the target’s next location. Only the SNs close to the predicted location remain active for the next sensing iteration.

In contrast, dynamic clustering forms clusters only when the necessity arises or on the onset of some triggered event. A relevant work in Zhang et al.²⁸ proposed an adaptive consensus-based distributed filter (ACDF) scheme to track linearly moving targets. The scheme dynamically performs hierarchical clustering to calculate target location using the odometry motion model. Here, the CH is selected based on residual energy followed by cluster formation. The SNs within the cluster forward target location to the CH, which uses Kalman filtering (KF) to calculate the target location. This is followed by the addition and removal of SNs from the cluster based on the moving target. Since only SNs within the cluster gets activated, this results in prolonged WSN lifetime. However, the scheme fails to cater to random target path changes with no recovery mechanisms. Generally, in such scenarios, a more centralized approach²⁹ is used. That is, in addition to the CH responsible for the aggregation of target location information from the SNs, it forwards this aggregated information to the BS. The BS maintains a target failure probability (TFP) against each CH, in case one is lost, the node with the smallest TFP is selected as the new CH.

A dynamic clustering-based energy efficient target tracking, called Improved Prediction-based Adaptive-Head (IPAH), algorithm has been proposed in Darabkh et al.³⁰ They couple prediction method with the proposed scheme to get better results. IPAH efficiently selects CH and efficiently tracks the target.

In summary, clustering makes the network flexible and scalable. It provides the topology for accurate target tracking while ensuring efficient bandwidth utilization and data transmission rate.

Non-clustering-based target tracking

In non-clustering-based network topology, each SN contributes equally in network development and deployment. The topology is set to achieve best effort delivery augmented with probability-based target tracking techniques to reduce duplicate packet flow and data flooding. A relevant study in Luo et al.³² proposes a cooperative localization and tracking algorithm (CLTA) for an indoor environment. The algorithm is divided into two phases: offline and online. In the offline phase, the network is divided into grid-based known and/or reliable SNs. These nodes subsequently take part in target localization. During the online phase, the target location is calculated using a centroid weighting algorithm for finding overlapping regions. However, the technique fails at grid borders. Moreover, the overall network energy depletes much faster because the SNs are active all the time irrespective of the target location.

In Bhowmik and Giri,³³ a tree-based localization scheme is proposed, referred to as convoy tree-based fuzzy target tracking (CTFTT). The scheme incorporates fuzzy sensing model (FSM) with RSSI to track the target. Once an SN detects the target in its vicinity, it gets connected to the tree and starts following the target within a monitoring area. The scheme results in maximal coverage with farther SNs remaining asleep and, hence, saving energy. Nonetheless, in dense networks, the overall energy dissipation is higher with SNs joining and leaving the tree frequently. Moreover, in the case of a fast-moving target, the scheme fails to locate the target accurately. There exist more robust techniques similar to one proposed in Yu³⁵ that integrates the concept of data association to determine the most suitable set of SNs for target localization.

Even though, non-clustering-based schemes require no topology maintenance, they are easy to scale out with low packet overhead. On the contrary, clustering-based schemes incur more packet overhead during the CH selection. Thereafter, the schemes provide accurate target localization and better network lifetime.

Target tracking in IIoT

With the advancement of IIoT, the communication between operators and equipment referred to as cobots has improved. With robust techniques for tracking, real-time data forwarding makes it a cornerstone for industrial automation. Traditionally, the communication between operators is directly linked to the assembly line, a slow process. Moreover, the lack of tracking affects production, especially in assembly lines manufacturing multiple goods.

Over the years, many solutions are developed for the purpose of tracking on the factory floor. One such attempt is OmniTrack,³⁶ an orientation-aware radio-frequency identification (RFID) tracking solution. The solution is based on signal phase changes to predict the target location and direction. The specialized and standalone hardware has its limitations; for instance, it only works unless the perpendicular distance from the antennas to the target location is known. Similar RFID-based localization system³⁷ locates and tracks the target in a complex environment. It adopts a double tag for RFID localization to minimize the negative effect of propagation and eradicate the phase ambiguity. The result is greater accuracy when localizing the target.

More sophisticated mechanisms, like one in Chiani et al.,³⁸ implement a non-cooperative target tracking framework for an indoor environment using impulse radio (IR) and ultra-wideband (UWB) technology. A radar sensor network (RSN) coupled with signal processing steps is used to achieve high localization accuracy. Though the RSN with no prior knowledge of target location and/or its trajectory generates a high number of false alarms. This can be improved by distinguishing between true and false alarms, that is, using a ghost mitigation algorithm or a naive sample wise detector with a constant threshold to improve network performance. Moreover, the framework incorporates cardinalized probability hypothesis density (CPHD) filter to further improve target detection and tracking accuracy. However, this adds up to additional computational complexity, affecting the overall network lifetime.

More recent cooperative techniques³⁹ are developed to work with heterogeneous networks. Here, some of the nodes (RFIDs, cameras, and proximity sensors) are used to achieve high accuracy of target tracking, while other passive nodes are used to process, aggregate, and forward the information. In addition, some prediction mechanisms can be incorporated to predict the target next location. Undoubtedly, modern industrial services heavily rely on IIoT to locate and track workers and industrial equipment on the factory floor. There is a need to develop frameworks capable of accurate localization and tracking to ensure improved industrial productivity and safety. All this while ensuring a favorable balance between target localization accuracy and network energy depletion.

System model

Consider a set of SNs $S$ with limited data processing capacity, mainly due to the battery constraint. The nodes are placed uniformly in the area under surveillance. Each node $s \in S$ is equipped with an omnidirectional antenna and is only aware of the nodes within its transmission range. For communication between the SNs and the target is based on an ad hoc on-demand distance vector (AODV) routing protocol. Every deployment starts with the formation of a cluster $C$ with a CH $k$ and cluster id $Ω (k)$ . Here, $k$ is responsible for processing the received information from the cluster members. The result is an aggregated target location, which is forwarded to the BS.

It is worthwhile to mention that tracking a moving target using deployed sensors can become inefficient due to the limited power available. In many of the existing techniques either the SN energy depletes rapidly or it compromises the tracking accuracy. There is a need for a robust tracking scheme that maintains a balance between efficiency and accuracy.

Design and implementation

CH selection for IIoT environment

The CH selection is important for efficient management of data flow in IIoT environment. Here, we have proposed an adaptive CH selection mechanism. The method selects the CH at runtime as all the sensors are not active, that is to save energy. Once the SNs $s \in S$ detect the target in its range, it informs its immediate neighbors. Only the nodes close to the target awake while others remain asleep. This results in preserving the overall network energy, subsequently prolonging the network lifetime. Moreover, the awake SNs form a cluster $C$ and select a CH $k$ that acts as a mediator between the SNs and the BS. Note that $k$ is selected based on its distance from the BS. The CH selection process is listed in Algorithm 1.

Algorithm 1 Cluster head selection
Input $C$ : cluster of alive sensor nodes; $BS$ : base stationOutput $k$ : cluster head1: ${k \| k \in C_{Ω (i)}^{k}, Ω (k) = Ω (i)}$ ▹ Cluster head already elected2: if $k \neq ϕ$ then3: return $k$ 4: end if5: return ${k \| \min (d_{s, B}, S_{id}), s \in S}$ ▹ Elect one nearest to the $BS$

The CH selection is a two-phase process. During the initialization phase, the SN detects the target in its range. All the SNs $s \in S$ compute their distances from the BS. The one with the shortest distance becomes the candidate node for becoming the CH $k$ . In the case of multiple candidates, the distance and energy and node id are used to tie break among the SNs. The node having smallest id is selected as a CH. Since the target is generally moving in an IIoT environment, the CH selection is repeated as shown in Figure 1.

Figure 1.

Flowchart of the proposed dynamic clustering-based target tracking scheme.

Real-time Target Tracking in IIoT

In IIoT, sensors are deployed in a specific region to track the movement of cobots and equipment as shown in Figure 2. Due to limited battery power, it is unfeasible to keep the SNs in surveillance mode all the time. This results in rapid energy depletion. Thus, to optimize the use of energy, we propose a dynamic clustering-based multilateral target tracking framework. The framework allows the SNs to change their states from active to sleep and vice versa. This change in state is managed dynamically through context sensing. Each node at a random point in time becomes active to sense its surrounding, in the case of no anomalies, it goes back to sleep state. This mechanism is used to initially detect the target of $t$ . This mechanism is repeated unless $t$ is detected. When detected, nearby nodes become active $S_{A}$ and start tracking the target location, whereas the remaining nodes can move to sleep state $S - S_{A}$ . Furthermore, the active nodes inform the neighboring nodes about the target movement. This collaborative mechanism allows suitable SNs to become active and start tracking the target as listed in Algorithm 2.

Figure 2.

Typical sensors network topology.

Algorithm 2 Target tracking after regular intervals.
Input $S$ : set of sensors; $BS$ : base station; $T$ : timerOutput $t$ : target location1: set_status( $\forall s \in S$ , ASLEEP)2: while true do3: if T EXPIRED then4: for each $s \in S$ do5: if sense $(s)$ $\neq ϕ$ then ▹ Target within sensor range6: set_status( $s$ , ALIVE)7: end if8: end for9: $C \leftarrow$ form({ $s \| s \in S$ , get_status $(s)$ = ALIVE}) ▹ Cluster formation10: $k \leftarrow$ elect( $C$ ) ▹ Cluster head selection11: for each $s \in C$ do12: send( $d_{s, t}, k$ ) ▹ Forward distance to target13: set_status( $s$ , ASLEEP)14: end for15: $t \leftarrow$ process() ▹ Compute target location at cluster head16: send( $t, BS$ ) ▹ Forward target location to $BS$ 17: reset( $T$ )18: end if19: end while

Algorithm 2 Target tracking after regular intervals.

Input

S

: set of sensors;

BS

: base station;

T

: timerOutput

t

: target location1: set_status(

\forall s \in S

, ASLEEP)2: while true do3: if T EXPIRED then4: for each

s \in S

do5: if sense

(s)

\neq ϕ

then ▹ Target within sensor range6: set_status(

s

, ALIVE)7: end if8: end for9:

C \leftarrow

form({

s | s \in S

, get_status

(s)

= ALIVE}) ▹ Cluster formation10:

k \leftarrow

elect(

C

) ▹ Cluster head selection11: for each

s \in C

do12: send(

d_{s, t}, k

) ▹ Forward distance to target13: set_status(

s

, ASLEEP)14: end for15:

t \leftarrow

process() ▹ Compute target location at cluster head16: send(

t, BS

) ▹ Forward target location to

BS

17: reset(

T

)18: end if19: end while

CH processing

The active nodes $S_{A}$ track the movement of the target and share real-time information with the CH $k$ . The $k$ removes any redundancies and aggregate all the received information. Once processed, $k$ computes the target location and sends it to the respective BS for analysis and decision making. Note that the proposed scenario provides real-time target tracking or anomaly detection mechanism in an industrial environment.

Performance evaluation

The proposed algorithm is designed to improve the accuracy and reliability of a mobile target in the IIoT environment. The simulation environment consists of the uniformly distributed SNs in an industrial environment within the area of 100 × 100 m². To verify the efficiency and accuracy of our proposed scheme, we simulate the target tracking using two approaches: regular interval and complete path. First, we evaluate the tracking accuracy of our proposed scheme under a different number of rounds (time steps). Second, we simulate the proposed scheme for complete path tracking of a cobot or equipment in an industrial environment. To do this, we simulate three different tracks separately and compare the results. The target tracks are randomly generated using normal distribution represented as

$χ ~ N (μ, σ^{2})$ (1)

where $χ$ is a random variable distributed normally with mean $μ$ and variance $σ^{2}$ , that is, it is used to get random points on x-axis and y-axis.

Furthermore, the probability density function (PDF) of the target $t$ is computed as

$f (t | μ, σ^{2}) = \frac{1}{σ} φ (\frac{t - μ}{σ})$ (2)

where $φ (t) = (1 / \sqrt{2 π}) e^{- \frac{1}{2} t^{2}}$ shows the probability of target with a maximum value of $1 / \sqrt{2 π}$ .

The target distance $d_{s, t}$ from the SNs $s \in S$ is calculated as

$d_{s, t} = \sqrt{{(t_{x} - s_{x})}^{2} + {(t_{y} - s_{y})}^{2}}$ (3)

where $(t_{x}, t_{y})$ is the target location and $(s_{x}, s_{y})$ is the SN location. This distance is used to awake SNs in the vicinity of the target being tracked.

For evaluation, we compare both schemes in terms of average error per meter denoted as $ϵ$ using

$ϵ = \sqrt{{(x - x')}^{2} + {(y - y')}^{2}}$ (4)

where $(x, y)$ is the predicted target location, whereas $(x', y')$ is the actual target location.

Regular interval

In this scenario, the target has been tracked after regular interval of time $T$ . Each SN senses the target in its range. If the target is within range, they form a cluster and elects a CH, otherwise, the SN goes back to sleep state for a specific period $T$ .

We evaluate the proposed algorithm with varying number of rounds to identify the target locations after every round. For all the experiments, the round interval is set to 5 s. The comparison is performed with the trilateration technique proposed in Khan et al.²⁷ We plot the output of target tracking simulation time against root mean square error (RMSE) generated in tracking to observe the behavior of the proposed scheme under regular interval as well as complete path tracking. Furthermore, we plot the output as a cumulative distribution function (CDF) against tracking RMSE, to get the probability of tracking error in the proposed and the trilateration technique.

Figure 3 shows the tracking RMSE error overtime with varying round intervals when tracking a target. Note that the proposed scheme localizes the target after regular intervals referred to as rounds. We observe that the proposed scheme performs better localization of the target overtime compared to trilateration. This is due to the fact that trilateration is limited to positioning information from only the first three sensing nodes, not the case in the proposed dynamic clustering scheme with more information for reliable target tracking. Moreover, the average target localization errors at different rounds are summarized in Table 2. The results show that the proposed scheme predicts the location of the target node with better accuracy. Moreover, with the increasing number of moving steps, the proposed scheme outperforms the trilateration technique in terms of positioning accuracy of tracking.

Figure 3.

Localization positioning error overtime in terms of RMSE (in meters) for the proposed technique compared to trilateration with varying round intervals (5, 10, and 15 s).

Table 2.

Comparison of average localization positioning error in terms of RMSE (in meters).

Number of rounds	10	20	30	40
Trilateration	2.024	1.655	1.929	1.877
Proposed	1.318	1.292	1.455	1.503

RMSE: root mean square error.

In Figure 4, we plot the CDF of the tracking error measured as RMSE to evaluate the performance of the proposed scheme against the trilateration scheme. We observe that the overall performance of the proposed scheme is better than trilateration. Moreover, we test this observation using the Kolmogorov–Smirnov test with the hypothesis that CDF of trilateration lies below that of CDF (proposed) in the case of round interval 5 s. The probability that CDF of trilateration is worse than CDF of proposed is 33% (p-value = 0.33). On the contrary, the probability that CDF of trilateration is worse than CDF of proposed stands at 100% (p-value = 1.0) for round intervals 10 and 15 s. A detailed comparison of the probabilities is shown in Table 3.

Figure 4.

Cumulative distribution function plot for localization positioning error (RMSE in meters) of the proposed technique compared to trilateration with varying round intervals (5, 10, and 15 s) from left to right.

Table 3.

Localization positioning error CDF in terms of RMSE in meters.

		RMSE CDF
		0.1	0.5	0.9
Trilateration	50 s	0.493	2.025	3.556
	100 s	0.785	1.656	3.234
	150 s	1.033	1.929	3.555
	200 s	0.979	1.877	3.509
Proposed	50 s	0.141	1.317	3.453
	100 s	0.432	1.292	2.852
	150 s	0.487	1.454	3.209
	200 s	0.533	1.503	3.264

RMSE: root mean square error; CDF: cumulative distribution function.

The behavior of the proposed technique is further tested in different sensor density areas like $8 \times 8$ , $16 \times 16$ , and $32 \times 32$ . In Figure 5, we plot the positioning error against varying sensor densities and sensing interval rounds. The results show that the proposed scheme outperforms the traditional trilateration scheme for target tracking.

Figure 5.

Box plots for localization positioning error in terms of RMSE (in meters) of the proposed technique compared to trilateration with varying sensor densities (8 × 8, 16 × 16, and 32 × 32) for different round intervals (5, 10, and 15 s for subplots from top to bottom).

Complete path

In an industrial environment, with the lack of prior information, unpredictable movement of the target, and inefficient tracking systems can affect the overall automation system. Therefore, three different target mobility models are selected based on their path complexity level and commonly used in WSN. The selected models can give a better comparison between the proposed and conventional trilateration models. In this scenario, each SN senses the target in its range. If the target is within the range they form a cluster $C$ and select a CH $k$ . The cluster $C$ tracks the target location and forwards it to the $BS$ after processing and removing redundant data. For evaluation, we use different mobility models like spiral, random walk, and Manhattan to demonstrate the aforementioned environment.

The results of target tracking under different mobility models are shown in Figure 6. The simulation results show that the proposed scheme provided 22% performance improvement under the spiral model, whereas 17% and 18% performance is achieved in random walk and Manhattan models, respectively. The performance gain is achieved in comparison with the base trilateration scheme. Table 4 summarizes the results of the proposed scheme against the base scheme.

Figure 6.

Target tracking for three different target mobility models.

Table 4.

Comparison of average tracking error in terms of RMSE (in meters) for different mobility models.

Mobility model	Average RMSE		Performance
	Trilateration	Proposed	Gain (%)
Spiral	1.662	1.289	22
Random walk	1.851	1.539	17
Manhattan	1.899	1.558	18

RMSE: root mean square error.

Discussion

IIoT has a strategic value for smart factories soon. It improves the overall efficiency of internal operations while reducing the overall operational cost and also reflects in the production of low cost and affordable products. Therefore, the use of smart equipment, especially cobots on factory floors, informing about their status has become an important selling point for smart factories. It not only enables the different equipment on the factory floor to communicate in real-time with one another but also improves the overall manufacturing process with more control. For instance, failure of equipment may not get reported on time, affecting the overall performance of other systems. Such failures are important to detect and track at the earliest.

No doubt tracking of such faulty equipment in a real-time environment is challenging for industrial applications where accuracy and run-time performance are important factors. In this article, we presented a robust target/fault tracking scheme that is capable of overcoming some of the problems faced in industrial applications. The solution can be deployed in areas where human access is difficult like rotatory machinery, unstrapped vehicles, steel melting units, and radiation units. Therefore, we proposed this solution to support real-time monitoring and fault detection.

Nevertheless, the proposed scheme can also be used for rescue operations. In such situations, the trustworthiness of the received information makes the difference between life and death. For instance, when workers are trapped in underground units due to earthquake, flood, explosion, or any other natural disaster. The proposed scheme can pinpoint the precise location of the workers through effective deployment of Internet of things (IoT) resources.

Conclusion

In this article, we propose a cooperative multilateral sensing scheme for target tracking in an industrial environment. We demonstrate that the proposed scheme improves the accuracy and reliability of target tracking. A simulation with regular interval tracking and complete path tracking has been carried out to analyze the proposed scheme. Our results show that the proposed scheme improves the accuracy and reliability of target tracking, especially in an indoor industrial environment. In the future, we would like to implement and investigate our scheme for an outdoor environment. Such environments span over a vast area with challenges of coverage and network lifetime. Moreover, an extension for seamless integration of indoor and outdoor environments will help to predict and track the target more accurately and precisely. In IIoT environment, CH selection should be based on the distance between SNs and the BS. This becomes challenging when different factors like energy, reliability of link, and distance between competing SNs affect the performance. Nevertheless, in the future, we would also like to implement and extend the scheme while maintaining a balance between these factors.

Footnotes

Handling Editor: SooKyun Kim

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research,authorship,and/or publication of this article.

Funding

The author(s) received no financial support for the research,authorship,and/or publication of this article.

ORCID iDs

Anis U Rahman

Asad Waqar Malik

References

Patwari

. Location Estimation in Sensor Networks. Pennsylvania, PA: Citeseer, 2005.

Nakandala

Namasivayam

Chandima

, et al. Detecting wild elephants via WSN for early warning system. In: Proceedings of the 7th international conference on information and automation for sustainability, pp.1–6, Colombo, Sri Lanka, 22–24 December 2014. New York: IEEE.

Stattner

Vidot

Hunel

, et al. Wireless sensor network for habitat monitoring: a counting heuristic. In: Proceedings of the 37th Annual IEEE conference on local computer networks-workshops, pp.753–760, Clearwater, FL, 22–25 October 2012. New York: IEEE.

Kassim

MRM

Mat

Harun

. Wireless sensor network in precision agriculture application. In: Proceedings of the 2014 international conference on computer, information and telecommunication systems (CITS), pp.1–5, Jeju, South Korea, 7–9 July 2014. New York: IEEE.

Dai

Wang

Yan

. BSHM-WSN: A wireless sensor network for bridge structure health monitoring. In: Proceedings of the 2012 international conference on modelling, identification and control, pp.708–712, Wuhan, China, 24–26 June 2012. New York: IEEE.

Savić

Radonjić

. WSN architecture for smart irrigation system. In: Proceedings of the 2018 23rd international scientific-professional conference on information technology (IT), pp.1–4, Zabljak, 19–24 February 2018. New York: IEEE.

Harun

Kassim

MRM

Mat

, et al. Precision irrigation using wireless sensor network. In: Proceedings of the 2015 international conference on smart sensors and application (ICSSA), pp.71–75, Kuala Lumpur, Malaysia, 25–28 May 2015. New York: IEEE.

Thekkil

Prabakaran

. Real-time WSN based early flood detection and control monitoring system. In: Proceedings of the 2017 international conference on intelligent computing, instrumentation and control technologies (ICICICT), pp.1709–1713, Kannur, India, 6–7 July 2017. New York: IEEE.

Ghaleb

Felemban

Subramaniam

, et al. A performance simulation tool for the analysis of data gathering in both terrestrial and underwater sensor networks. IEEE Access 2017; 5: 4190–4208.

10.

Erol-Kantarci

Mouftah

Oktug

. A survey of architectures and localization techniques for underwater acoustic sensor networks. IEEE Commun Surv Tutor 2011; 13(3): 487–502.

11.

Liu

Chen

Lin

, et al. Developed urban air quality monitoring system based on wireless sensor networks. In: Proceedings of the 2011 fifth international conference on sensing technology, pp.549–554, Palmerstone North, 28 November–1 December 2011. New York: IEEE.

12.

Saboor

Mustafa

Ahmad

, et al. Evolution of wireless standards for health monitoring. In: Proceedings of the information technology, electromechanical and microelectronics conference (IEMECON), Jaipur, India, 13–15 March 2019. New York: IEEE.

13.

Zubiete

Luque

Rodríguez

AVM

, et al. Review of wireless sensors networks in health applications. Conf Proc IEEE Eng Med Biol Soc 2011; 2011: 1789–1793.

14.

Wei

Deng

. A cluster-growing dimension of complex networks: from the view of node closeness centrality. Phys A: Stat Mech Appl 2019; 522: 80–87.

15.

Yang

Deng

Jones

. Network division method based on cellular growth and physarum-inspired network adaptation. Int J Unconv Comput 2018; 13(6): 477–491.

16.

Pant

Verma

Dhuliya

. A study on disaster detection and management using WSN in Himalayan region of Uttarakhand. In: Proceedings of the 2017 3rd international conference on advances in computing, communication & automation (ICACCA) (Fall), pp.1–6, Dehradun, India, 15–16 September 2017.

17.

Singh

Saha

Chugh

, et al. Distributed event detection in wireless sensor networks for forest fires. In: Proceedings of the 2013 UKSim 15th international conference on computer modelling and simulation, pp.634–639, Cambridge, 10–12 April 2013. New York: IEEE.

18.

Parthasarathy

Whiton

Hagerskans

, et al. An in-vehicle wireless sensor network for heavy vehicles. In: Proceedings of the 2016 IEEE 21st international conference on emerging technologies and factory automation (ETFA), pp.1–8, Berlin, 6–9 September 2016. New York: IEEE.

19.

Faschinger

Sastry

Patel

, et al. An RFID and wireless sensor network-based implementation of workflow optimization. In: Proceedings of the 2007 IEEE international symposium on a world of wireless, mobile and multimedia networks, pp.1–8, Espoo, 18–21 June 2007. New York: IEEE.

20.

Kandhare

Bhandari

. Guidance providing navigation in target tracking for wireless sensor networks. IJSR 2015; 4: 2795–2798.

21.

Deshpande

Grant

Henderson

. Target localization and autonomous navigation using wireless sensor networks—a pseudogradient algorithm approach. IEEE Syst J 2013; 8(1): 93–103.

22.

Avgousti

Panayides

Christoforou

, et al. Medical telerobotics and the remote ultrasonography paradigm over 4g wireless networks. In: Proceedings of the 2018 IEEE 20th international conference on e-Health networking, applications and services (Healthcom), pp.1–6, Ostrava, 17–20 September 2018. New York: IEEE.

23.

Hamdan

Shanableh

Zaki

, et al. IoT-based interactive dual mode smart home automation. In: Proceedings of the 2019 IEEE international conference on consumer electronics (ICCE), pp.1–2, Las Vegas, NV, 11–13 January 2019. New York: IEEE.

24.

Thapliyal

Ratajczak

Wendroth

, et al. Amazon echo enabled IoT home security system for smart home environment. In: Proceedings of the 2018 IEEE international symposium on smart electronic systems (iSES) (Formerly iNiS), pp.31–36, Hyderabad, India, 17–19 December 2018. New York: IEEE.

25.

Gubbi

Buyya

Marusic

, et al. Internet of things (IoT): a vision, architectural elements, and future directions. Future Gener Comput Syst 2013; 29(7): 1645–1660.

26.

Yang

Shen

Wang

. The internet of things in manufacturing: key issues and potential applications. IEEE Syst Man Cyb Mag 2018; 4(1): 6–15.

27.

Khan

Saqib

Khan

. Target tracking in wireless sensor networks using NS2. In: Yang

Nagar

Joshi

(eds) Smart trends in systems, security and sustainability. New York: Springer, 2018, pp.21–31.

28.

Zhang

Zhou

Wang

, et al. Adaptive consensus-based distributed target tracking with dynamic cluster in sensor networks. IEEE T Cybernet 2018; 49(5): 1580–1591.

29.

Khakpour

Pazzi

El-Khatib

. Using clustering for target tracking in vehicular ad hoc networks. Veh Commun 2017; 9: 83–96.

30.

Darabkh

Albtoush

Jafar

. Improved clustering algorithms for target tracking in wireless sensor networks. The Journal of Supercomputing 2017; 73(5): 1952–1977.

31.

Rouhani

Haghighat

. Boundary static clustering target tracking in wirless sensor networks. In: Proceedings of the 2015 6th international conference on computing, communication and networking technologies (ICCCNT), pp.1–6, Denton, TX, 13–15 July 2015. New York: IEEE.

32.

Luo

Zhang

Liu

, et al. Reliable and cooperative target tracking based on WSN and Wi-Fi in indoor wireless networks. IEEE Access 2018; 6: 24846–24855.

33.

Bhowmik

Giri

. Convoy tree based fuzzy target tracking in wireless sensor network. Int J Wirel Inform Netw 2017; 24(4): 476–484.

34.

Abid

Khan

Hayat

, et al. Real-time object tracking in wireless sensor network. In: Proceedings of the 2017 10th international conference on electrical and electronics engineering (ELECO), pp.1103–1107, Bursa, 30 November–2 December 2017. New York: IEEE.

35.

. Distributed target tracking in wireless sensor networks with data association uncertainty. IEEE Commun Lett 2017; 21(6): 1281–1284.

36.

Jiang

Zheng

, et al. Orientation-aware RFID tracking with centimeter-level accuracy. In: Proceedings of the 2018 17th ACM/IEEE international conference on information processing in sensor networks (IPSN), pp.290–301, Porto, 11–13 April 2018. New York: IEEE.

37.

Xiao

Wang

, et al. One more tag enables fine-grained RFID localization and tracking. IEEE/ACM T Netw 2018; 26(1): 161–174.

38.

Chiani

Giorgetti

Paolini

. Sensor radar for object tracking. Proc IEEE 2018; 106(6): 1022–1041.

39.

Liu

Zhao

. Collaborative target tracking of iot heterogeneous nodes. Measurement 2019; 147: 106872.

Exploiting cooperative sensing for accurate target tracking in industrial Internet of things

Abstract

Keywords

Introduction

Problem statement

Related work

Clustering-based target tracking

Non-clustering-based target tracking

Target tracking in IIoT

System model

Design and implementation

CH selection for IIoT environment

Real-time Target Tracking in IIoT

CH processing

Performance evaluation

Regular interval

Complete path

Discussion

Conclusion

Footnotes

Declaration of conflicting interests

Funding

ORCID iDs

References