Area Coverage Optimization in Wireless Sensor Network by

Semi-random Deployment

Adda Boualem

, Youcef Dahmani

, Abdelkader Maatoug

and Cyril De-runz

The High National School of Computer Science, ESI, Algiers, Algeria

The EECE Lab, Ibn Khaldoun University, Tiaret, Algeria

The CReSTIC, University of Reims Champagne-Ardenne, Reims, France

Keywords:

Semi-random Deployment, Area Coverage, Power Consumption, Wireless Sensor Network.

Abstract:

Area Coverage is a difﬁcult problem to solve with a minimum number of sensor nodes for maximum time

duration, especially in wide areas. Among the solutions proposed for this problem there is the deployment

with its two types random and deterministic. The disadvantages of deterministic deployment are conﬁgured in

the Area of Interest (AoI) limitation, and random deployment is conﬁgured in the non-equitable distribution

of sensor nodes on AoI. This problem affects power consumption and connectivity; as a result, it affects the

coverage of the area of interest. In this paper we have proposed a third type of hybrid deployment that gathers

the advantages and minimizes the disadvantage of the application of the random deployment and deterministic

deployment. This type has two steps; the anticipate conﬁguration step and the scheduling process step. A

comparative study was done to show the effectiveness of this type to optimize coverage in the vast area of

interest less danger and more interests.

1 INTRODUCTION

The advancement in Micro Electro-Mechanical Sys-

tems (MEMS) technology has lead to production of

thin, cheaper and reliable sensors nodes. A network

of sensors suffers from several technical weak points

such as communication range, monitoring range, low

battery, and network deployment circumstance prob-

lems such as the difﬁculty of building a sensor net-

work in volcanoes, mountains, or in the oceans. The

deployment with its two types considered a very ef-

fective solution to optimize the problem of coverage

in networks of wireless sensors, power consumption,

routing, connectivity, and others. Deterministic de-

ployment is effective in very limited areas where man

can intervene to troubleshoot, conﬁgure, change bat-

teries, and replace or move nodes. Random deploy-

ment is effective in unrestricted areas where human

intervention is impossible or difﬁcult. Advancement

in MEMS technology has lead to production of very

thin sensor nodes that are less expensive and yet pow-

erful in terms of storage and processing. Increasing

network lifetime by keeping the entire coverage of the

area of interest is one of the compromises desired by

researchers in their work using deployment. Among

the disadvantages of this solution for the random type

are the non-equitable distribution of the sensor nodes

and the difﬁculty of applying the process of activating

sensor nodes in the network. The necessity of know-

ing positions (best distribution) and remote-control in

geographical areas is mandatory, where human inter-

vention is difﬁcult in certain danger zones. In this

paper, we have deﬁned a new type of deployment

known as the semi-random deployment. This deploy-

ment uses deterministic advantages like knowing the

distribution of positions. It also employs random de-

ployment mechanisms of dispersing nodes randomly.

The semi-random deployment protocol consists of

two steps i.e, the anticipate conﬁguration step and the

scheduling process step. The anticipate conﬁguration

step consists of the conﬁguration of certain necessary

parameters and storing them in the memory only once

for the used in the nodes scheduling process. Our sim-

ulations showthat, the semi-random deploymentis ef-

ﬁcient compared to existing methods of deployment.

The rest of this paper is organized as follows. Section

2 formally deﬁnes the problem. Sections 3 reviews

some related work. Sections 4 proposes our sensor

deployment type. Simulation results are presented in

Section 5. The Results Analysis are discussed in Sec-

tion 6. Finally, the Conclusion is drawn in Section

Boualem, A., Dahmani, Y., Maatoug, A. and De-runz, C.

Area Coverage Optimization in Wireless Sensor Network by Semi-random Deployment.

DOI: 10.5220/0006581900850090

In Proceedings of the 7th International Conference on Sensor Networks (SENSORNETS 2018), pages 85-90

ISBN: 978-989-758-284-4

2 RELATED WORK

Currently, the use of wireless sensor networks is

a major need in various surveillance applications,

such as monitoring applications in; the environment,

battleﬁelds, borders, forests, intelligent spaces, and

in the control of industrial and biological disasters.

The main objective of using of the wireless sensors

network is how to monitor with sufﬁcient coverage

an area of interest with a fewer sensor nodes in a

longtime (cardie and Wu, 2004), (C.-F and Tseng,

2005).The guarantee of the perfect coverage leads

to guarantee the connectivity of the area of inter-

est (AoI). Coverage is difﬁcult to control, measured

and guaranteed using a minimal number of sensors

nodes deployed on AoI; so, coverage is an optimiza-

tion problem. Deployment is one of the solutions

proposed to solve this problem. Deterministic de-

ployment is preferred in areas where human inter-

vention is possible to; replace, or to change sensor

nodes. On the other hand; random deployment is nec-

essary in the terrain that represents the danger and

the impossibility of the intervention of the human be-

ing. The treatment of the deployment-based cover-

age problem is extensively addressed using different

domains strategies as: using the scheduling process

between the Active / Passive states of the nodes in

the network (M. Zaied, 2003), using the geometric

angles and distances between the nodes (Aurenham-

mer, 2001), using the disjoint dominating setsand grid

strategies (Shen and Sun, 2006), using the Voronoi

diagram and the Delaunay triangulation (Aurenham-

mer, 2001), using the heuristics (M. Zaied, 2003), and

others technics . (Wen-Hwa Liao, 2001) Presents the

Glowworm Swarm Optimization (GSO) protocol; a

new system process based on random deployment to

improve of the nodes coverage. GSO protocol con-

siders each node as a single glowworms transmitter

and luminant substance like ”luciferin”. The luciferin

force considered as the link between the transmit-

ter node and its neighboring sensors. A sensor node

move to the low-density area if necessary. A cover-

age maximization was achieved when a sensor node

can move to the low-density area. The disadvantage

of GSM is that the nodes must provide by mobilizers

and a GPS position detection system, which quickly

depletes the network. (X. M. Guo, 2012) divides the

area of interest into grid and the selected sensor node

in the next round is there situated in the best case.

The goal behind this method is to select a smallest

set of nodes to guarantee the target coverage and de-

termine the precise positions for the deployed sensor

nodes in the network. This method is not efﬁcient

in the area coverage and in the wide area where the

monitoring of each point is necessary. (Alduraibi Fa-

had and Younis, 2016) considers the coverage as an

optimization problem, where they dress the problem

with the deployment of a smallest set of nodes to

maximize the coverage. The authors propose three

models optimization models. The ﬁrst is minimiz-

ing the sensor nodes number deployed into the area

of interest to attain the high reliability level detec-

tion. The second model based on the determination

of available nodes positions with nodes number con-

strain for attain perfect coverage. The third optimiza-

tion model is to minimize the nodes number deployed

in some locations that require low coverage and ad-

just the nodes number deployed in others locations

that require higher coverage. (O. Banimelhem, 2013)

the implementation of a genetic algorithm (GA). This

algorithm speciﬁes the positions and the number of

sensor nodes to be deployed in the zone of interest

in order to achieve optimal coverage. The applica-

tion of this algorithm overloads the nodes by calcula-

tions that always leave the nodes in active state, the

thing that quickly exhausts the network. The remedy

proposed for this is the repetitive deployment of sen-

sor nodes in locations where coverage has reached a

minimal threshold, which is not always possible es-

pecially in danger zones. In (Shen and Sun, 2006),

the different strategies are used to maximize cover-

age in WSN. These strategies are classiﬁed into three

categories: (a) Force Based, (b) Grid Based, and

ployment strategies depend on the mobility of sen-

sor nodes, where; Sensor nodes are forced to move

away or approach each other until they reach the per-

fect coverage. The Force Based strategy uses a vir-

tual force as a repugnant and attractive force. The

study in (Howard and Sukhatme, 2002) considers

that AoI contains sensor nodes and objects that ex-

ert a virtual repulsive force to move the sensor nodes

away from the objects and also from each other so

that their surveillance areas do not overlap. The sen-

sor nodes will continue to move until they reach the

steady state in AoI. This strategy provides coverage

and connectivity on AoI, but it is extremely depen-

dent on mobility and complex computations, which

quickly depletes the network. The grid-based deploy-

ment strategy consists of determining the precise po-

sitions of the sensors. This is the strategy in which the

coverage is evaluated as the ratio between the points

of the covered grid and the total number of points of

the grid in the AoI. The accuracy of the assessment

is determined by the size of each grid, the smallest

in size is the most accurate in the assessment. (Shen

and Sun, 2006) mentions three types of grids used

in the deployment to provide the best coverage; (A)

SENSORNETS 2018 - 7th International Conference on Sensor Networks

Triangular grid, (b) Square grid, and (c) Hexagonal

grid. The triangular strategy is the best type because

it is the smallest area of overlap, so fewer sensors,

then the other types (Sanjay Shakkottai, 2003), and

the hexagonal grid is the worst of all, since it has

the largest area Overlap. Computational geometry is

frequently used to optimize coverage in WSN, (Au-

renhammer, 2001) mentions two geometric computa-

tional approaches used in deployment to provide the

best coverage; (A) Voronoi diagram, and (b) Delau-

nay triangulation. The Voronoi diagram is the divi-

sion of the area of interest on locations (sites) so that

the points in a polygon are closer to the site in the

polygon than the other sites. The Voronoi diagram

can be used as a sampling method to determine cov-

erage, and measure the holes of in WSN; the sensors

act as sites, if all vertices of the Voronoi polygons are

covered, then the AoI is fully covered, otherwise the

coverage holes exist. The Delaunay triangle is formed

by three sites provided if and only if the sites (Wang

and Porta, 2003) circulate do not contain other sites.

Delaunay triangulation coverage strategy: it is fre-

quently used in coverage optimization in WSN. The

Delaunay triangulation is used to evaluate the worst

and best case of coverage (Megerian and Srivastava,

2005). The complexity of these approaches lays in the

use of computational complexity geometry methods,

which quickly depletes the network.

In (Faten Hajjej and Zaied, 2016), a Flower Polli-

nation Algorithm (FPA) was introduced as a solution

for deployment and coverage problem. A deployment

approach based on FPA was proposed for the sensor

placement problem for WSN. This approach can ﬁnd

the optimal placement topology in terms of Quality of

service (QoS) metric. But FPA uses too many param-

eters to calculate for specify the positions of active

nodes, which quickly depletes the network.

3 ASSUMPTIONS

To apply our strategy of area coverage in large areas,

which consists of using semi-random deployment by

mixing between the two types (random and determin-

istic), we took into consideration the following as-

sumptions:

1. The subdivision of AoI under sub-areas, in order

to cover the vast areas well, to achieve an equi-

table distribution of the nodes in the geographical

area, to control connectivity, and to conserve en-

ergy consumption in the network.

2. The deployment of a large number of sensor nodes

on each sub-area (sub-zone) by a device on a

height a little lower and in calm circumstances

around each of the sub-zones z

, in order to guar-

antee a good distribution in the network.

3. The nodes used in our study are of two types; the

ﬁrst type is made to have a storage memory more

than the ordinal nodes, the second type has the

same characteristics of the ﬁrst but a little heavy

than the nodes of the ﬁrst type.

4 DESCRIPTION OF OUR

APPROACH

Our protocol consists of two main steps, the anticipate

conﬁguration step and the scheduling process step.

Figure 1: (a) one sub-area, and (b) subdivision of AoI in n

sub-areas.

4.1 Anticipate Conﬁguration Step

The area of interest must be divided geographically

by n sub-areas of dimension L×L and diagonal D, as

shown in Figure1. Anticipate conﬁguration consists

of the following steps:

1. The communication radius R

, and monitoring ra-

dius R

have the following features:

= R

(1)

√

2×L (2)

We have taken the case where the communication

radius equals the surveillance radius, as shown

in relation to equation (1). For a single sen-

sor node to be able to fully cover the sub-area

that belongs to it and communicate the collected

data to the base station, the communication ra-

dius and the monitoring radius would have to be

1.42×(thesub−areaside) , As the relation (2).

2. The nodes used in this study are homogeneous

and endowed with a memory capable of storing

the following parameters:

Area Coverage Optimization in Wireless Sensor Network by Semi-random Deployment

• T

act

: Contains the duration of activation or the

time of one revolution.

• T

att

: Contains the time to wait for the conﬁrma-

tion message by the node that will be active in

the next round. The time T

att

, T

act

(the waiting

time is less than the activation time (the time of

one round)) .T

att

act

times are chosen accord-

ing to the conﬁdentiality of the task to be per-

formed and the danger of the zone.

• Num: Contains the number of the node, this pa-

rameter is unique for each node, varied by 1..n.

For each sub-area.

• Num −zone: Contains the number of the sub-

zone (sub-area) to be deployed.

• Each node N

has the ability to generate a ran-

dom number ρ

• A special node Nsp

(sink

), it was manufac-

tured for each sub-zone z

, must be heavier

compared to the other nodes of z

, to fall around

the center at the time of deployment. This node

must be programmed to be active at the time of

their arrival on z

to allow selecting their neigh-

boring nodes by sending a message Hello. The

other objective of this sink is the communica-

tion of all data collected to the main base sta-

tion (BS).

4.2 Scheduling Process Step

The scheduling process step consists of the following

steps:

1. All nodes are passive after deployment except the

sinks (special nodes).

2. The special node Nsp

(sink

) sends a message

Hello contains the parameter Num −zone to the

neighboring nodes, and aims to know the nodes

that belong to its own sub-zone. The recipient

nodes update their old Num −zone (ie become

neighbors of the new Nsp

3. In the ﬁrst round, the node Nsp

generates a ran-

dom number ρ

and determines the next active

node in its own subarea to act as the coverage

and transmission of the data harvested according

to the following formula:

= [ρ

×n

] (3)

Such as:

• The generate number ρ

/ ρ

∈]0..1[ , and

• n

represents the total number of nodes de-

ployed in the appropriate sub-zone z

• [ρ

×n

] represents the integer part of the calcu-

lated number n

×n

4. The node selected to be active in the next round

must activate their transmitter and respond with

a conﬁrmation message to the transmitter node (a

Hello message contains the number of the trans-

mitter node). If the transmitter node has not re-

ceived the conﬁrmation message in a time T

att

, the

latter will remain active until the ﬁnal exhaustion

of their battery.

5 SIMULATION

We compared this type of deployment with the pro-

tocols that use deployment and the Active / Passive

scheduling process in this case study. The data used

in this evaluation are recorded in Table1.

To compare our protocol called Semi-Random De-

ployment Protocol (SRDP) with the protocol named

Flower Pollination Coverage Optimization approach

(FPCOA) cited in (Faten Hajjej and Zaied, 2016),

and mount our improvements over it; we took the

same simulation data cited in (Faten Hajjej and Zaied,

2016).The deployment of a number of sensor nodes

between 5 and 40; and an area of interest of dimen-

sions 50m 50m; to measure the coverage ratio, and

thus to calculate the energyconsumed duringthe 1000

registered primary time units.

To apply our protocol, we divide the area of inter-

est into L-side of sub-areas such that: L = R

/1.42 =

15/1.42 = 10.56m ie: 25sub −areas. The results of

simulation are shown in Figure2 and Figure3.

Table 1: Common data used in the simulations.

Conﬁgurations Value

Area of Interest (AoI) 50m×50m

Communication Radius (R

) 15m

Sensing Radius (R

) 15m

Deployed Sensor Nodes Number 5to40

Time Units 1000units

Figure 2: The Comparison of Coverage Ratio with SRDP

and FPCOA protocols.

SENSORNETS 2018 - 7th International Conference on Sensor Networks

Figure 3: The Comparison of total energy consumption with

SRDP and FPCOA protocols.

6 RESULTS ANALYSIS

We simulated the total energy consumption data, thus

the coverage ratio in the network for both protocols,

the Semi-Random Deployment Protocol (SRDP) and

the Flower Pollination Coverage Optimization Ap-

proach (FPCOA).

6.1 In Terms of Coverage

The SRDP protocol arrives at a ratio of 1 if the num-

ber of nodes deployed reaches 25. On the other hand,

the coverage ratio by applying FPCOA arrives at 1 if

the number of nodes arrives at 35 nodes as shown in

Figure2; Which shows a gain at the minimum sensor

node number that can be deployed to achieve a per-

fect coverage of the area of interest. In addition, the

network keeps a perfect coverage lifetime more than

applies the protocol FPCOA; Because SRDP uses

the clustering concept with the activation of a single

Cluster-Head by a Cluster to act as a monitoring and

communication role of the data captured at the base

station.

6.2 In Terms of Energy Consumption

Figure3 shows an equitable energy consumption in

the network by the application of the SRDP, modeled

and proved by the function Y = 500−X. On the other

hand, the graph representing the energy consumption

of the FPCOA isa graph of the fractions; which shows

unfair energy consumption in the network. Figure3

shows that the network is exhausted during the 500

prime time units if by applying the FPCOA; On the

other hand the network consumes half of their total

energy in this time, and it happens to consume their

total energy during the 1000 units of prime time. This

shows the effectiveness of SRDP versus FPCOA.

6.3 The Advantages of this Strategy

Compared to the Previous

Scheduling Strategies

1. Not interested in the number of turns for the ap-

plication of the active / passive process made by

each node in the network.

2. Not interested in implementing a process to select

the nodes of each subﬁeld (of each cluster) that

overloads the node processor and wastes energy

and time.

3. Not interested in knowing geographical position,

as a result do not need a location system like GPS

except in the initial deployment step (anticipate

conﬁguration step).

4. Activate a single node in each sub-zone to bal-

ance the network in terms of power consumption,

guarantee connectivity, and increase the network

lifetime.

5. Activation of a single node per sub-zone guaran-

tees full coverage of the area of interest for maxi-

mum time without having to implement strategies

for measuring coverage.

6. Each asset plays a dual role; a monitoring role and

a communication role.

7. The compromise between the two types of de-

ployment brings beneﬁts and minimizes their dis-

advantages.

7 CONCLUSIONS

In this paper, we proposed a new type of deployment

based on anticipate conﬁguration that serves to ini-

tialize some parameters help to balance energy con-

sumption, offer connectivity and serves in the entire

coverage of the area of interest, thus helps to avoid the

different scheduling process and clustering processes

used in the sensor network implementation without

affecting the connectivity. The subdivision of the area

of interest in small areas of diameter equal to the mon-

itoring radius R

, and putting R

= R

provides fair re-

source consumption in the network, thus guaranteeing

connectivity in AoI throughout the network lifetime.

The generation of a random number by each node

helps to select the next active node for each subﬁeld

without the need for different active / passive mecha-

nisms used for this purpose. Using memory settings

helps to avoid too much computation and to save en-

ergy. The coverage considering R

= R

is shown

to be effective in area coverage when we subdivide

Area Coverage Optimization in Wireless Sensor Network by Semi-random Deployment

the AoI in subareas of diameter D, (R

= R

= D), in

contrast in previous protocols proposed in (Faten Haj-

jej and Zaied, 2016), (Yang, 2012).The only inconve-

nient is the use of memory parameters that it will be

processed in the upcoming works.

REFERENCES

Alduraibi Fahad, N. L. and Younis, M. (2016). Coverage-

based node placement optimization in wireless sen-

sor network with linear topology. In (ICC), C., editor,

IEEE International Conference on. IEEE. IEEE Inter-

national Conference on. IEEE.

Aurenhammer, F. (2001). Computational geometry : Some

easy questions and their recent solutions. Journal of

Universal Computer Science, 7(5).

C.-F, H. and Tseng, Y.-C. (2005). A survey of solutions

to the coverage problems in wireless sensor networks,

journal of internet technology. Journal of internet

Technology, 6(1):1–8.

cardie, M. and Wu (2004). Coverage problem in Wireless

Ad Hoc Sensor Networks, Handbook of Sensor Net-

works, chapter 19. CRC Press.

Faten Hajjej, R. E. and Zaied, M. (2016). An efﬁcient

deployment approach for improved coverage in wire-

less sensor networks based on ﬂower pollination algo-

rithm. pages 117– 129. Computer Science and Infor-

mation Technology (CS and IT).

Howard, A., M. M. and Sukhatme (2002). Mobile sen-

sor network deployment using potential ﬁelds: A

distributed, scalable solution to the area coverage

problem. In 6th International Symposium on Dis-

tributed Autonomous Robotics Systems, pages 299–

308, Fukuoka, Japan.

M. Zaied, C. Ben Amar, M. A. A. (2003). Award a new

wavelet based beta function. In International con-

ference on signal, system and design, SSD03 1, vol-

ume 1, pages 185–191.

Megerian, S., K. F. P. M. and Srivastava, M. (2005). Worst

and best-case coverage in sensor networks. In IEEE

Transactions on Mobile Computing, volume 1, pages

84 – 92. IEEE CS, CASS, ComSoc, IES, and SPS.

O. Banimelhem, M. Mowaﬁ, W. A. (2013). Genetic algo-

rithm based deployment in hybrid wireless sensor net-

works. Communications and Network.

Sanjay Shakkottai, R. Srikant, N. S. (2003). Unreliable sen-

sor grids: Coverage, connectivity and diameter. In

Twenty-Second Annual Joint Conference of the IEEE

Computer and Communications Societies, volume 2,

pages 1073–1083, Austin TX, USA.

Shen, X., C. J. W. Z. and Sun, Y. (2006). Grid scan: A sim-

ple and effective approach for coverage issue in wire-

less sensor networks. Istanbul, Turkey. Communica-

tions, 2006. ICC 06. IEEE International Conference.

Wang, G., C. G. and Porta, T. (2003). A bidding protocol for

deploying mobile sensors network protocols. In 11th

IEEE International Conference on Network Protocols,

pages 315 – 324.

Wen-Hwa Liao, Yucheng Kao, Y.-S. L. (2001). A sensor

deployment approach using glowworm swarm opti-

mization algorithm in wireless sensor networks. EL-

SEVIER Journal, Expert Systems with Applications,

38(10):12180–12188.

X. M. Guo, C. J. Zhao, X. T. Y. e. a. (2012). A deterministic

sensor node deployment method with target coverage

based on grid scan. Chinese Journal of Sensors and

Actuators, 25(1):104–109.

Yang, X.-S. (2012). Flower pollination algorithm for global

optimization. In on Unconventional Computing, I. C.

and Computation, N., editors, Flower Pollination Al-

gorithm for Global Optimization, volume 7445, pages

240–249.

SENSORNETS 2018 - 7th International Conference on Sensor Networks