1 \documentclass[a4paper,twoside]{article}
13 \usepackage{SCITEPRESS}
14 \usepackage[small]{caption}
16 \usepackage[linesnumbered,ruled,vlined,commentsnumbered]{algorithm2e}
17 \usepackage{mathtools}
25 %\title{Authors' Instructions \subtitle{Preparation of Camera-Ready Contributions to SCITEPRESS Proceedings} }
27 \title{Distributed Lifetime Coverage Optimization Protocol \\in Wireless Sensor Networks}
29 \author{\authorname{Ali Kadhum Idrees, Karine Deschinkel, Michel Salomon, and Rapha\"el Couturier}
30 \affiliation{FEMTO-ST Institute, UMR 6174 CNRS, University of Franche-Comte, Belfort, France}
31 %\affiliation{\sup{2}Department of Computing, Main University, MySecondTown, MyCountry}
32 \email{ali.idness@edu.univ-fcomte.fr, $\lbrace$karine.deschinkel, michel.salomon, raphael.couturier$\rbrace$@univ-fcomte.fr}
33 %\email{\{f\_author, s\_author\}@ips.xyz.edu, t\_author@dc.mu.edu}
36 \keywords{Wireless Sensor Networks, Area Coverage, Network lifetime,
37 Optimization, Scheduling.}
39 \abstract{ One of the main research challenges faced in Wireless Sensor Networks
40 (WSNs) is to preserve continuously and effectively the coverage of an area (or
41 region) of interest to be monitored, while simultaneously preventing as much
42 as possible a network failure due to battery-depleted nodes. In this paper we
43 propose a protocol, called Distributed Lifetime Coverage Optimization protocol
44 (DiLCO), which maintains the coverage and improves the lifetime of a wireless
45 sensor network. As a first step we partition the area of interest into
46 subregions using a classical divide-and-conquer method. Our DiLCO protocol is
47 then distributed on the sensor nodes in each subregion in a second step. To
48 fulfill our objective, the proposed protocol combines two effective
49 techniques: a leader election in each subregion, followed by an
50 optimization-based node activity scheduling performed by each elected leader.
51 This two-step process takes place periodically, in order to choose a small set
52 of nodes remaining active for sensing during a time slot. Each set is built
53 to ensure coverage at a low energy cost, allowing to optimize the network
54 lifetime. More precisely, a period consists of four phases: (i)~Information
55 Exchange, (ii)~Leader Election, (iii)~Decision, and (iv)~Sensing. The
56 decision process, which result in an activity scheduling vector, is carried
57 out by a leader node through the solving of an integer program. In comparison
58 with some other protocols, the simulations done using the discrete event
59 simulator OMNeT++ show that our approach is able to increase the WSN lifetime
60 and provides improved coverage performance. }
62 \onecolumn \maketitle \normalsize \vfill
64 \section{\uppercase{Introduction}}
65 \label{sec:introduction}
67 Energy efficiency is a crucial issue in wireless sensor networks since sensory
68 consumption, in order to maximize the network lifetime, represent the major
69 difficulty when designing WSNs. As a consequence, one of the scientific research
70 challenges in WSNs, which has been addressed by a large amount of literature
71 during the last few years, is the design of energy efficient approaches for
72 coverage and connectivity~\cite{conti2014mobile}. Coverage reflects how well a
73 sensor field is monitored. The most discussed coverage problems in literature
74 can be classified into three types \cite{li2013survey}: area coverage (where
75 every point inside an area is to be monitored), target coverage (where the main
76 objective is to cover only a finite number of discrete points called targets),
77 and barrier coverage (to prevent intruders from entering into the region of
78 interest). On the one hand we want to monitor the area of interest in the most
79 efficient way~\cite{Nayak04}. On the other hand we want to use as less energy as
80 possible. Sensor nodes are battery-powered with no means of recharging or
81 replacing, usually due to environmental (hostile or unpractical environments) or
82 cost reasons. Therefore, it is desired that the WSNs are deployed with high
83 densities so as to exploit the overlapping sensing regions of some sensor nodes
84 to save energy by turning off some of them during the sensing phase to prolong
87 In this paper we design a protocol that focuses on the area coverage problem
88 with the objective of maximizing the network lifetime. Our proposition, the
89 DiLCO protocol, maintains the coverage and improves the lifetime in WSNs. The
90 area of interest is first divided into subregions using a divide-and-conquer
91 algorithm and an activity scheduling for sensor nodes is then planned by the
92 elected leader in each subregion. In fact, the nodes in a subregion can be seen
93 as a cluster where each node sends sensing data to the cluster head or the sink
94 node. Furthermore, the activities in a subregion/cluster can continue even if
95 another cluster stops due to too many node failures. Our Distributed Lifetime
96 Coverage Optimization (DILCO) protocol considers periods, where a period starts
97 with a discovery phase to exchange information between sensors of a same
98 subregion, in order to choose in a suitable manner a sensor node (the leader) to
99 carry out the coverage strategy. In each subregion the activation of the sensors
100 for the sensing phase of the current period is obtained by solving an integer
103 The remainder of the paper continues with Section~\ref{sec:Literature Review}
104 where a review of some related works is presented. The next section describes
105 the DiLCO protocol, followed in Section~\ref{cp} by the coverage model
106 formulation which is used to schedule the activation of
107 sensors. Section~\ref{sec:Simulation Results and Analysis} shows the simulation
108 results. The paper ends with conclusions and some suggestions for further work
109 in Section~\ref{sec:Conclusion and Future Works}.
111 \section{\uppercase{Literature Review}}
112 \label{sec:Literature Review}
113 \noindent In this section, we summarize some related works regarding coverage lifetime maximization and scheduling, and distinguish our DiLCO protocol from the works presented in the literature. Some algorithms have been developed in ~\cite{yang2014energy,ChinhVu,vashistha2007energy,deschinkel2012column,shi2009,qu2013distributed,ling2009energy,xin2009area,cheng2014achieving,ling2009energy} to solve the area coverage problem so as to preserve coverage and prolong the network lifetime.
116 Yang et al.~\cite{yang2014energy} investigated full area coverage problem
117 under the probabilistic sensing model in the sensor networks. They have studied the relationship between the
118 coverage of two adjacent points mathematically and then convert the problem of full area coverage into point coverage problem. They proposed $\varepsilon$-full area coverage optimization (FCO) algorithm to select a subset
119 of sensors to provide probabilistic area coverage dynamically so as to extend the network lifetime.
122 Vu et al.~\cite{ChinhVu} proposed a localized and distributed greedy algorithm named DESK for generating non-disjoint cover sets which provide the k-area coverage for the whole network.
125 Qu et al.~\cite{qu2013distributed} developed a distributed algorithm using adjustable sensing sensors
126 for maintaining the full coverage of such sensor networks. The
127 algorithm contains two major parts: the first part aims at
128 providing $100\%$ coverage and the second part aims at saving
129 energy by decreasing the sensing radius.
131 Shi et al.~\cite{shi2009} modeled the Area Coverage Problem (ACP), which will be changed into a set coverage
132 problem. By using this model, they are proposed an Energy-Efficient central-Scheduling greedy algorithm, which can reduces energy consumption and increases network lifetime, by selecting a appropriate subset of sensor nodes to support the networks periodically.
134 The work in~\cite{cheng2014achieving} presented a unified sensing architecture for duty cycled sensor networks, called uSense, which comprises three ideas: Asymmetric Architecture, Generic Switching and Global Scheduling. The objective is to provide a flexible and efficient coverage in sensor networks.
136 In~\cite{ling2009energy}, the lifetime of
137 a sensor node is divided into epochs. At each epoch, the
138 base station deduces the current sensing coverage requirement
139 from application or user request. It then applies the heuristic algorithm in order to produce the set of active nodes which take the mission of sensing during the current epoch. After that, the produced schedule is sent to the sensor nodes in the network.
144 The work in ~\cite{vu2009delaunay} considered the area coverage problem for variable sensing radii in WSNs by improving the energy balancing heuristic proposed in ~\cite{wang2007energy} so that the area of interest can be full covered using Delaunay triangulation structure.
146 Diongue and Thiare~\cite{diongue2013alarm} proposed an energy aware sleep scheduling algorithm for lifetime maximization in wireless sensor networks (ALARM). The proposed approach permits to schedule redundant nodes according to the weibull distribution. This work did not analyze the ALARM scheme under the coverage problem.
149 In~\cite{xin2009area}, the authors proposed a circle intersection localized coverage algorithm
150 to maintain connectivity based on loose connectivity critical condition
151 . By using the connected coverage node set, it can maintain network
152 connection in the case which loose condition is not meet.
153 The authors in ~\cite{vashistha2007energy} addressed the full area coverage problem using information
154 coverage. They are proposed a low-complexity heuristic algorithm to obtain full area information covers (FAIC), which they refer to as Grid Based FAIC (GB-FAIC) algorithm. Using these FAICs, they are obtained the optimal schedule for applying the sensing activity of sensor nodes in order to
155 achieve increased sensing lifetime of the network.
162 In \cite{xu2001geography}, Xu et al. proposed a Geographical Adaptive Fidelity (GAF) algorithm, which uses geographic location information to divide the area of interest into fixed square grids. Within each grid, it keeps only one node staying awake to take the responsibility of sensing and communication.
164 The main contributions of our DiLCO Protocol can be summarized as follows:
165 (1) The distributed optimization over the subregions in the area of interest,
166 (2) The distributed dynamic leader election at each round by each sensor node in the subregion,
167 (3) The primary point coverage model to represent each sensor node in the network,
168 (4) The activity scheduling based optimization on the subregion, which are based on the primary point coverage model to activate as less number as possible of sensor nodes to take the mission of the coverage in each subregion, and (5) The improved energy consumption model.
171 The work presented in~\cite{luo2014parameterized,tian2014distributed} tries to solve the target coverage problem so as to extend the network lifetime since it is easy to verify the coverage status of discreet target.
172 %Je ne comprends pas la phrase ci-dessus
173 The work proposed in~\cite{kim2013maximum} considers the barrier-coverage problem in WSNs. The final goal is to maximize the network lifetime such that any penetration of the intruder is detected.
174 %inutile de parler de ce papier car il concerne barrier coverage
175 In \cite{ChinhVu}, the authors propose a localized and distributed greedy algorithm named DESK for generating non-disjoint cover sets which provide the k-coverage for the whole network.
176 Our Work in~\cite{idrees2014coverage} proposes a coverage optimization protocol to improve the lifetime in heterogeneous energy wireless sensor networks. In this work, the coverage protocol distributed in each sensor node in the subregion but the optimization take place over the the whole subregion. We are considered only distributing the coverage protocol over two subregions.
178 The work presented in ~\cite{Zhang} focuses on a distributed clustering method, which aims to extend the network lifetime, while the coverage is ensured.
180 The work proposed by \cite{qu2013distributed} considers the coverage problem in WSNs where each sensor has variable sensing radius. The final objective is to maximize the network coverage lifetime in WSNs.
184 Casta{\~n}o et al.~\cite{castano2013column} proposed a multilevel approach based on column generation (CG) to extend the network lifetime with connectivity and coverage constraints. They are included two heuristic methods within the CG framework so as to accelerate the solution process.
185 In \cite{diongue2013alarm}, diongue is proposed an energy Aware sLeep scheduling AlgoRithm for lifetime maximization in WSNs (ALARM) algorithm for coverage lifetime maximization in wireless sensor networks. ALARM is sensor node scheduling approach for lifetime maximization in WSNs in which it schedule redundant nodes according to the weibull distribution taking into consideration frequent nodes failure.
186 Yu et al.~\cite{yu2013cwsc} presented a connected k-coverage working sets construction
187 approach (CWSC) to maintain k-coverage and connectivity. This approach try to select the minimum number of connected sensor nodes that can provide k-coverage ($k \geq 1$).
188 In~\cite{cheng2014achieving}, the authors are presented a unified sensing architecture for duty cycled sensor networks, called uSense, which comprises three ideas: Asymmetric Architecture, Generic Switching and Global Scheduling. The objective is to provide a flexible and efficient coverage in sensor networks.
190 In~\cite{yang2013energy}, the authors are investigated full area coverage problem
191 under the probabilistic sensing model in the sensor networks. %They are designed $\varepsilon-$full area coverage optimization (FCO) algorithm to select a subset of sensors to provide probabilistic area coverage dynamically so as to extend the network lifetime.
192 In \cite{xu2001geography}, Xu et al. proposed a Geographical Adaptive Fidelity (GAF) algorithm, which uses geographic location information to divide the area of interest into fixed square grids. Within each grid, it keeps only one node staying awake to take the responsibility of sensing and communication.
194 The main contributions of our DiLCO Protocol can be summarized as follows:
195 (1) The distributed optimization over the subregions in the area of interest,
196 (2) The distributed dynamic leader election at each round by each sensor node in the subregion,
197 (3) The primary point coverage model to represent each sensor node in the network,
198 (4) The activity scheduling based optimization on the subregion, which are based on the primary point coverage model to activate as less number as possible of sensor nodes to take the mission of the coverage in each subregion,
199 (5) The improved energy consumption model.
203 \section{ The DiLCO Protocol Description}
204 \label{sec:The DiLCO Protocol Description}
206 \noindent In this section, we introduce the DiLCO protocol which is distributed
207 on each subregion in the area of interest. It is based on two efficient
208 techniques: network leader election and sensor activity scheduling for coverage
209 preservation and energy conservation, applied periodically to efficiently
210 maximize the lifetime in the network.
211 \iffalse The main features of our DiLCO protocol: i)It divides the area of
212 interest into subregions by using divide-and-conquer concept, ii)It requires
213 only the information of the nodes within the subregion, iii) it divides the
214 network lifetime into rounds, iv)It based on the autonomous distributed decision
215 by the nodes in the subregion to elect the Leader, v)It apply the activity
216 scheduling based optimization on the subregion, vi) it achieves an energy
217 consumption balancing among the nodes in the subregion by selecting different
218 nodes as a leader during the network lifetime, vii) It uses the optimization to
219 select the best representative set of sensors in the subregion by optimize the
220 coverage and the lifetime over the area of interest, viii)It uses our proposed
221 primary point coverage model, which represent the sensing range of the sensor as
222 a set of points, which are used by the our optimization algorithm, ix) It uses a
223 simple energy model that takes communication, sensing and computation energy
224 consumptions into account to evaluate the performance of our protocol.
227 \subsection{ Assumptions and models}
229 \noindent We consider a sensor network composed of static nodes distributed
230 independently and uniformly at random. A high density deployment ensures a high
231 coverage ratio of the interested area at the starting. The nodes are supposed to
232 have homogeneous characteristics from a communication and a processing point of
233 view, whereas they have heterogeneous energy provisions. Each node has access
234 to its location thanks, either to a hardware component (like a GPS unit), or a
235 location discovery algorithm.
237 \indent We consider a boolean disk coverage model which is the most widely used
238 sensor coverage model in the literature. Thus, since a sensor has a constant
239 sensing range $R_s$, every space points within a disk centered at a sensor with
240 the radius of the sensing range is said to be covered by this sensor. We also
241 assume that the communication range $R_c \geq 2R_s$. In fact, Zhang and
242 Zhou~\cite{Zhang05} proved that if the transmission range fulfills the previous
243 hypothesis, a complete coverage of a convex area implies connectivity among the
244 working nodes in the active mode.
246 \indent For each sensor we also define a set of points called primary
247 points~\cite{idrees2014coverage} to approximate the area coverage it provides,
248 rather than working with a continuous coverage. Thus, a sensing disk
249 corresponding to a sensor node is covered by its neighboring nodes if all its
250 primary points are covered. Obviously, the approximation of coverage is more or
251 less accurate according to the number of primary points.
254 By knowing the position (point center: ($p_x,p_y$)) of a wireless
255 sensor node and its $R_s$, we calculate the primary points directly
256 based on the proposed model. We use these primary points (that can be
257 increased or decreased if necessary) as references to ensure that the
258 monitored region of interest is covered by the selected set of
259 sensors, instead of using all the points in the area.
261 \indent We can calculate the positions of the selected primary
262 points in the circle disk of the sensing range of a wireless sensor
263 node (see figure~\ref{fig1}) as follows:\\
264 $(p_x,p_y)$ = point center of wireless sensor node\\
266 $X_2=( p_x + R_s * (1), p_y + R_s * (0) )$\\
267 $X_3=( p_x + R_s * (-1), p_y + R_s * (0)) $\\
268 $X_4=( p_x + R_s * (0), p_y + R_s * (1) )$\\
269 $X_5=( p_x + R_s * (0), p_y + R_s * (-1 )) $\\
270 $X_6= ( p_x + R_s * (\frac{-\sqrt{2}}{2}), p_y + R_s * (0)) $\\
271 $X_7=( p_x + R_s * (\frac{\sqrt{2}}{2}), p_y + R_s * (0))$\\
272 $X_8=( p_x + R_s * (\frac{-\sqrt{2}}{2}), p_y + R_s * (\frac{-\sqrt{2}}{2})) $\\
273 $X_9=( p_x + R_s * (\frac{\sqrt{2}}{2}), p_y + R_s * (\frac{-\sqrt{2}}{2})) $\\
274 $X_{10}=( p_x + R_s * (\frac{-\sqrt{2}}{2}), p_y + R_s * (\frac{\sqrt{2}}{2})) $\\
275 $X_{11}=( p_x + R_s * (\frac{\sqrt{2}}{2}), p_y + R_s * (\frac{\sqrt{2}}{2})) $\\
276 $X_{12}=( p_x + R_s * (0), p_y + R_s * (\frac{\sqrt{2}}{2})) $\\
277 $X_{13}=( p_x + R_s * (0), p_y + R_s * (\frac{-\sqrt{2}}{2})) $.
283 %\includegraphics[scale=0.20]{fig21.pdf}\\~ ~ ~ ~ ~(a)
284 %\includegraphics[scale=0.20]{fig22.pdf}\\~ ~ ~ ~ ~(b)
285 \includegraphics[scale=0.25]{principles13.pdf}%\\~ ~ ~ ~ ~(c)
286 %\includegraphics[scale=0.10]{fig25.pdf}\\~ ~ ~(d)
287 %\includegraphics[scale=0.10]{fig26.pdf}\\~ ~ ~(e)
288 %\includegraphics[scale=0.10]{fig27.pdf}\\~ ~ ~(f)
290 \caption{Wireless Sensor Node represented by 13 primary points}
291 %\caption{Wireless Sensor Node represented by (a)5, (b)9 and (c)13 primary points respectively}
297 \subsection{The main idea}
300 \noindent We start by applying a divide-and-conquer algorithm to partition the
301 area of interest into smaller areas called subregions and then our protocol is
302 executed simultaneously in each subregion.
306 \includegraphics[width=75mm]{FirstModel.pdf} % 70mm
307 \caption{DiLCO protocol}
311 As shown in Figure~\ref{fig2}, the proposed DiLCO protocol is a periodic
312 protocol where each period is decomposed into 4~phases: Information Exchange,
313 Leader Election , Decision, and Sensing. For each period there will be exactly
314 one cover set in charge of the sensing task. A periodic scheduling is
315 interesting because it enhances the robustness of the network against node
316 failures. First, a node that has not enough energy to complete a period, or
317 which fails before the decision is taken, will be excluded from the scheduling
318 process. Second, if a node fails later, whereas it was supposed to sense the
319 region of interest, it will only affect the quality of coverage until the
320 definition of a new cover set in the next period. Constraints, like the energy
321 consumption, can be easily taken into consideration since the sensors can update
322 and exchange their information during the first phase. Let us notice that the
323 phases before the sensing one (Information Exchange, Leader Election, and
324 Decision) are energy consuming for all the nodes, even nodes that will not be
325 retained by the leader to keep watch over the corresponding area.
327 During the excution of the DiLCO protocol, two kinds of packets will be used:
328 %\begin{enumerate}[(a)]
330 \item INFO packet: sent by each sensor node to all the nodes inside a same
331 subregion for information exchange.
332 \item ActiveSleep packet: sent by the leader to all the nodes in its subregion
333 to inform them to be stay Active or to go Sleep during the sensing phase.
336 and each sensor node will have five possible status in the network:
337 %\begin{enumerate}[(a)]
339 \item LISTENING: sensor is waiting for a decision (to be active or not);
340 \item COMPUTATION: sensor applies the optimization process as leader;
341 \item ACTIVE: sensor is active;
342 \item SLEEP: sensor is turned off;
343 \item COMMUNICATION: sensor is transmitting or receiving packet.
347 An outline of the protocol implementation is given by Algorithm~\ref{alg:DiLCO}
348 which describes the execution of a period by a node (denoted by $s_j$ for a
349 sensor node indexed by $j$). At the beginning a node checks whether it has
350 enough energy to stay active during the next sensing phase. If yes, it exchanges
351 information with all the other nodes belonging to the same subregion: it
352 collects from each node its position coordinates, remaining energy ($RE_j$), ID,
353 and the number of one-hop neighbors still alive. Once the first phase is
354 completed, the nodes of a subregion choose a leader to take the decision based
355 on the following criteria with decreasing importance: larger number of
356 neighbors, larger remaining energy, and then in case of equality, larger index.
357 After that, if the sensor node is leader, it will execute the integer program
358 algorithm (see Section~\ref{cp}) which provides a set of sensors planned to be
359 active in the next sensing phase. As leader, it will send an Active-Sleep packet
360 to each sensor in the same subregion to indicate it if it has to be active or
361 not. Alternately, if the sensor is not the leader, it will wait for the
362 Active-Sleep packet to know its state for the coming sensing phase.
365 \subsubsection{Information Exchange Phase}
367 Each sensor node $j$ sends its position, remaining energy $RE_j$, and
368 the number of neighbours $NBR_j$ to all wireless sensor nodes in
369 its subregion by using an INFO packet and then listens to the packets
370 sent from other nodes. After that, each node will have information
371 about all the sensor nodes in the subregion. In our model, the
372 remaining energy corresponds to the time that a sensor can live in the
375 \subsubsection{Leader Election Phase}
376 This step includes choosing the Wireless Sensor Node Leader (WSNL),
377 which will be responsible for executing the coverage algorithm. Each
378 subregion in the area of interest will select its own WSNL
379 independently for each round. All the sensor nodes cooperate to
380 select WSNL. The nodes in the same subregion will select the leader
381 based on the received information from all other nodes in the same
382 subregion. The selection criteria in order of priority are: larger
383 number of neighbours, larger remaining energy, and then in case of
384 equality, larger index.
386 \subsubsection{Decision phase}
387 The WSNL will solve an integer program (see section~\ref{cp}) to
388 select which sensors will be activated in the following sensing phase
389 to cover the subregion. WSNL will send Active-Sleep packet to each
390 sensor in the subregion based on the algorithm's results.
393 \subsubsection{Sensing phase}
395 Active sensors in the round will execute their sensing task to preserve maximal
396 coverage in the region of interest. We will assume that the cost of keeping a
397 node awake (or asleep) for sensing task is the same for all wireless sensor
398 nodes in the network. Each sensor will receive an Active-Sleep packet from WSNL
399 informing it to stay awake or to go to sleep for a time equal to the period of
400 sensing until starting a new round. Algorithm 1, which will be executed by each
401 node at the beginning of a round, explains how the Active-Sleep packet is
408 \subsection{DiLCO protocol Algorithm}
409 we first show the pseudo-code of DiLCO protocol, which is executed by each
410 sensor in the subregion and then describe it in more detail. \fi
412 \begin{algorithm}[h!]
413 % \KwIn{all the parameters related to information exchange}
414 % \KwOut{$winer-node$ (: the id of the winner sensor node, which is the leader of current round)}
416 %\emph{Initialize the sensor node and determine it's position and subregion} \;
418 \If{ $RE_j \geq E_{th}$ }{
419 \emph{$s_j.status$ = COMMUNICATION}\;
420 \emph{Send $INFO()$ packet to other nodes in the subregion}\;
421 \emph{Wait $INFO()$ packet from other nodes in the subregion}\;
422 %\emph{UPDATE $RE_j$ for every sent or received INFO Packet}\;
423 %\emph{ Collect information and construct the list L for all nodes in the subregion}\;
425 %\If{ the received INFO Packet = No. of nodes in it's subregion -1 }{
426 \emph{LeaderID = Leader election}\;
427 \If{$ s_j.ID = LeaderID $}{
428 \emph{$s_j.status$ = COMPUTATION}\;
429 \emph{$\left\{\left(X_{1},\dots,X_{k},\dots,X_{J}\right)\right\}$ =
430 Execute Integer Program Algorithm($J$)}\;
431 \emph{$s_j.status$ = COMMUNICATION}\;
432 \emph{Send $ActiveSleep()$ to each node $k$ in subregion} \;
433 \emph{Update $RE_j $}\;
436 \emph{$s_j.status$ = LISTENING}\;
437 \emph{Wait $ActiveSleep()$ packet from the Leader}\;
438 % \emph{After receiving Packet, Retrieve the schedule and the $T$ rounds}\;
439 \emph{Update $RE_j $}\;
443 \Else { Exclude $s_j$ from entering in the current sensing phase}
446 \caption{DiLCO($s_j$)}
452 The DiLCO protocol work in rounds and executed at each sensor node in the network , each sensor node can still sense data while being in
453 LISTENING mode. Thus, by entering the LISTENING mode at the beginning of each round,
454 sensor nodes still executing sensing task while participating in the leader election and decision phases. More specifically, The DiLCO protocol algorithm works as follow:
455 Initially, the sensor node check it's remaining energy in order to participate in the current round. Each sensor node determines it's position and it's subregion based Embedded GPS or Location Discovery Algorithm. After that, All the sensors collect position coordinates, current remaining energy, sensor node id, and the number of its one-hop live neighbors during the information exchange. It stores this information into a list L.
456 The sensor node enter in listening mode waiting to receive ActiveSleep packet from the leader to take the decision. Each sensor node will execute the Algorithm~1 to know who is the leader. After that, if the sensor node is leader, It will execute the integer program algorithm ( see section~\ref{cp}) to optimize the coverage and the lifetime in it's subregion. After the decision, the optimization approach will select the set of sensor nodes to take the mission of coverage during the sensing phase. The leader will send ActiveSleep packet to each sensor node in the subregion to inform him to it's status during the period of sensing, either Active or sleep until the starting of next round. Based on the decision, the leader as other nodes in subregion, either go to be active or go to be sleep during current sensing phase. the other nodes in the same subregion will stay in listening mode waiting the ActiveSleep packet from the leader. After finishing the time period for sensing, all the sensor nodes in the same subregion will start new round by executing the DiLCO protocol and the lifetime in the subregion will continue until all the sensor nodes are died or the network becomes disconnected in the subregion.
460 \section{Coverage problem formulation}
463 \indent Our model is based on the model proposed by \cite{pedraza2006} where the
464 objective is to find a maximum number of disjoint cover sets. To accomplish
465 this goal, the authors proposed an integer program which forces undercoverage
466 and overcoverage of targets to become minimal at the same time. They use binary
467 variables $x_{jl}$ to indicate if sensor $j$ belongs to cover set $l$. In our
468 model, we consider binary variable $X_{j}$ which determine the activation of
469 sensor $j$ in the sensing phase. We also consider primary points as targets.
470 The set of primary points is denoted by $P$ and the set of sensors by $J$.
472 \noindent Let $\alpha_{jp}$ denote the indicator function of whether the primary
473 point $p$ is covered, that is:
475 \alpha_{jp} = \left \{
477 1 & \mbox{if the primary point $p$ is covered} \\
478 & \mbox{by sensor node $j$}, \\
479 0 & \mbox{otherwise.}\\
483 The number of active sensors that cover the primary point $p$ can then be
484 computed by $\sum_{j \in J} \alpha_{jp} * X_{j}$ where:
488 1& \mbox{if sensor $j$ is active,} \\
489 0 & \mbox{otherwise.}\\
493 We define the Overcoverage variable $\Theta_{p}$ as:
495 \Theta_{p} = \left \{
497 0 & \mbox{if the primary point}\\
498 & \mbox{$p$ is not covered,}\\
499 \left( \sum_{j \in J} \alpha_{jp} * X_{j} \right)- 1 & \mbox{otherwise.}\\
503 \noindent More precisely, $\Theta_{p}$ represents the number of active
504 sensor nodes minus one that cover the primary point $p$.\\
505 The Undercoverage variable $U_{p}$ of the primary point $p$ is defined
510 1 &\mbox{if the primary point $p$ is not covered,} \\
511 0 & \mbox{otherwise.}\\
516 \noindent Our coverage optimization problem can then be formulated as follows:
517 \begin{equation} \label{eq:ip2r}
520 \min \sum_{p \in P} (w_{\theta} \Theta_{p} + w_{U} U_{p})&\\
521 \textrm{subject to :}&\\
522 \sum_{j \in J} \alpha_{jp} X_{j} - \Theta_{p}+ U_{p} =1, &\forall p \in P\\
524 %\sum_{t \in T} X_{j,t} \leq \frac{RE_j}{e_t} &\forall j \in J \\
526 \Theta_{p}\in \mathbb{N} , &\forall p \in P\\
527 U_{p} \in \{0,1\}, &\forall p \in P \\
528 X_{j} \in \{0,1\}, &\forall j \in J
534 \item $X_{j}$ : indicates whether or not the sensor $j$ is actively sensing (1
535 if yes and 0 if not);
536 \item $\Theta_{p}$ : {\it overcoverage}, the number of sensors minus one that
537 are covering the primary point $p$;
538 \item $U_{p}$ : {\it undercoverage}, indicates whether or not the primary point
539 $p$ is being covered (1 if not covered and 0 if covered).
542 The first group of constraints indicates that some primary point $p$ should be
543 covered by at least one sensor and, if it is not always the case, overcoverage
544 and undercoverage variables help balancing the restriction equations by taking
545 positive values. Two objectives can be noticed in our model. First, we limit the
546 overcoverage of primary points to activate as few sensors as possible. Second,
547 to avoid a lack of area monitoring in a subregion we minimize the
548 undercoverage. Both weights $w_\theta$ and $w_U$ must be carefully chosen in
549 order to guarantee that the maximum number of points are covered during each
552 \section{\uppercase{Protocol evaluation}}
553 \label{sec:Simulation Results and Analysis}
554 \noindent \subsection{Simulation framework}
556 To assess the performance of our DiLCO protocol, we have used the discrete
557 event simulator OMNeT++ \cite{varga} to run different series of simulations.
558 Table~\ref{table3} gives the chosen parameters setting.
561 \caption{Relevant parameters for network initializing.}
564 % used for centering table
566 % centered columns (4 columns)
568 %inserts double horizontal lines
569 Parameter & Value \\ [0.5ex]
571 %Case & Strategy (with Two Leaders) & Strategy (with One Leader) & Simple Heuristic \\ [0.5ex]
575 % inserts single horizontal line
576 Sensing Field & $(50 \times 25)~m^2 $ \\
577 % inserting body of the table
579 Nodes Number & 50, 100, 150, 200 and 250~nodes \\
581 Initial Energy & 500-700~joules \\
583 Sensing Period & 60 Minutes \\
584 $E_{th}$ & 36 Joules\\
588 % [1ex] adds vertical space
594 % is used to refer this table in the text
597 Simulations with five different node densities going from 50 to 250~nodes were
598 performed considering each time 25~randomly generated networks, to obtain
599 experimental results which are relevant. The nodes are deployed on a field of
600 interest of $(50 \times 25)~m^2 $ in such a way that they cover the field with a
603 We chose as energy consumption model the one proposed proposed by~\cite{ChinhVu}
604 and based on ~\cite{raghunathan2002energy} with slight modifications. The energy
605 consumed by the communications is added and the part relative to a variable
606 sensing range is removed. We also assume that the nodes have the characteristics
607 of the Medusa II sensor node platform \cite{raghunathan2002energy}. A sensor
608 node typically consists of four units: a MicroController Unit, an Atmels AVR
609 ATmega103L in case of Medusa II, to perform the computations; a communication
610 (radio) unit able to send and receive messages; a sensing unit to collect data;
611 a power supply which provides the energy consumed by node. Except the battery,
612 all the other unit can be be switched off to save energy according to the node
613 status. Table~\ref{table4} summarizes the energy consumed (in milliWatt per
614 second) by a node for each of its possible status.
617 \caption{Energy consumption model}
620 % used for centering table
622 \begin{tabular}{|c|c|c|c|c|}
623 % centered columns (4 columns)
625 %inserts double horizontal lines
626 Sensor status & MCU & Radio & Sensing & Power (mW) \\ [0.5ex]
628 % inserts single horizontal line
629 Listening & ON & ON & ON & 20.05 \\
630 % inserting body of the table
632 Active & ON & OFF & ON & 9.72 \\
634 Sleep & OFF & OFF & OFF & 0.02 \\
636 Computation & ON & ON & ON & 26.83 \\
638 %\multicolumn{4}{|c|}{Energy needed to send/receive a 1-bit} & 0.2575\\
644 % is used to refer this table in the text
647 Less influent energy consumption sources like when turning on the radio,
648 starting the sensor node, changing the status of a node, etc., will be neglected
649 for the sake of simplicity. Each node saves energy by switching off its radio
650 once it has received its decision status from the corresponding leader (it can
651 be itself). As explained previously in subsection~\ref{main_idea}, two kinds of
652 packets for communication are considered in our protocol: INFO packet and
653 ActiveSleep packet. To compute the energy needed by a node to transmit or
654 receive such packets, we use the equation giving the energy spent to send a
655 1-bit-content message defined in~\cite{raghunathan2002energy} (we assume
656 symmetric communication costs), and we set their respective size to 112 and
657 24~bits. The energy required to send or receive a 1-bit is equal to $0.2575 mW$.
659 Each node has an initial energy level, in Joules, which is randomly drawn in the
660 interval $[500-700]$. If it's energy provision reaches a value below
661 $E_{th}=36$~Joules, the minimum energy needed for a node to stay active during
662 one period, it will no more participate in the coverage task. This value has
663 been computed by multiplying the energy consumed in active state (9.72 mW) by
664 the time in second for one round (3600 seconds). According to the interval of
665 initial energy, a sensor may be active during at most 20 rounds.
667 In the simulations, we introduce the following performance metrics to evaluate
668 the efficiency of our approach:
670 %\begin{enumerate}[i)]
673 \item {{\bf Coverage Ratio (CR)}:} it measures how well the WSN is able to
674 observe the area of interest. In our case, we discretized the sensor field
675 as a regular grid, which yields the following equation to compute the
679 \mbox{CR}(\%) = \frac{\mbox{$n$}}{\mbox{$N$}} \times 100.
681 where $n$ is the number of covered grid points by active sensors of every
682 subregions during the current sensing phase and $N$ is total number of grid
683 points in the sensing field. In our simulations, we have a layout of $N = 51
684 \times 26 = 1326$ grid points.
685 %The accuracy of this method depends on the distance between grids. In our
686 %simulations, the sensing field has been divided into 50 by 25 grid points, which means
687 %there are $51 \times 26~ = ~ 1326$ points in total.
688 % Therefore, for our simulations, the error in the coverage calculation is less than ~ 1 $\% $.
692 \item{{\bf Number of Active Sensors Ratio(ASR)}:} It is important to have as few active nodes as possible in each round,
693 in order to minimize the communication overhead and maximize the
694 network lifetime. The Active Sensors Ratio is defined as follows:
697 \mbox{ASR}(\%) = \frac{\sum\limits_{r=1}^R \mbox{$A_r^t$}}{\mbox{$S$}} \times 100 .
699 Where: $A_r^t$ is the number of active sensors in the subregion $r$ during round $t$ in the current sensing phase, $S$ is the total number of sensors in the network, and $R$ is the total number of the subregions in the network.
703 \item {{\bf Network Lifetime}:} we define the network lifetime as the time until
704 the coverage ratio drops below a predefined threshold. We denote by
705 $Lifetime_{95}$ (respectively $Lifetime_{50}$) the amount of time during which
706 the network can satisfy an area coverage greater than $95\%$ (respectively
707 $50\%$). We assume that the sensor network can fulfill its task until all its
708 nodes have been drained of their energy or it becomes disconnected. Network
709 connectivity is crucial because an active sensor node without connectivity
710 towards a base station cannot transmit any information regarding an observed
711 event in the area that it monitors.
713 \item {{\bf Energy Consumption}:} Energy Consumption (EC) can be seen as the
714 total energy consumed by the sensors during $Lifetime_{95}$ or
715 $Lifetime_{50}$, divided by the number of periods. Formally, the computation
716 of EC can be expressed as follows:
719 \mbox{EC} = \frac{\sum\limits_{m=1}^{M} \left( E^{\mbox{com}}_m+E^{\mbox{list}}_m+E^{\mbox{comp}}_m + E^{a}_m+E^{s}_m \right)}{M},
724 %\mbox{EC} = \frac{\mbox{$\sum\limits_{d=1}^D E^c_d$}}{\mbox{$D$}} + \frac{\mbox{$\sum\limits_{d=1}^D %E^l_d$}}{\mbox{$D$}} + \frac{\mbox{$\sum\limits_{d=1}^D E^a_d$}}{\mbox{$D$}} + %\frac{\mbox{$\sum\limits_{d=1}^D E^s_d$}}{\mbox{$D$}}.
727 where $M$ corresponds to the number of periods. The total energy consumed by
728 the sensors (EC) comes through taking into consideration four main energy
729 factors. The first one , denoted $E^{\scriptsize \mbox{com}}_m$, represent the
730 energy consumption spent by all the nodes for wireless communications during
731 period $m$. $E^{\scriptsize \mbox{list}}_m$, the next factor, corresponds to
732 the energy consumed by the sensors in LISTENING status before receiving the
733 decision to go active or sleep in period $m$. $E^{\scriptsize \mbox{comp}}_m$
734 refers to the energy needed by all the leader nodes to solve the integer program
735 during a period. Finally, $E^a_{m}$ and $E^s_{m}$ indicate the energy consumed
736 by the whole network in the sensing phase (active and sleeping nodes).
739 \item {{\bf Execution Time}:} a sensor node has limited energy resources and computing power,
740 therefore it is important that the proposed algorithm has the shortest
741 possible execution time. The energy of a sensor node must be mainly
742 used for the sensing phase, not for the pre-sensing ones.
744 \item {{\bf Stopped simulation runs}:} A simulation
745 ends when the sensor network becomes
746 disconnected (some nodes are dead and are not able to send information to the base station). We report the number of simulations that are stopped due to network disconnections and for which round it occurs.
754 %\subsection{Performance Analysis for different subregions}
755 \subsection{Performance Analysis}
758 In this subsection, we first focus on the performance of our DiLCO protocol for
759 different numbers of subregions. We consider partitions of the WSN area into
760 $2$, $4$, $8$, $16$, and $32$ subregions. Thus the DiLCO protocol is declined in
761 five versions: DiLCO-2, DiLCO-4, DiLCO-8, DiLCO-16, and DiLCO-32. Simulations
762 without partitioning the area of interest, case which corresponds to a
763 centralized approach, are not presented because they require high execution
764 times to solve the integer program and therefore consume too much energy.
766 We compare our protocol to two other approaches. The first one, called DESK and
767 proposed by ~\cite{ChinhVu} is a fully distributed coverage algorithm. The
768 second one, called GAF ~\cite{xu2001geography}, consists in dividing the region
769 into fixed squares. During the decision phase, in each square, one sensor is
770 chosen to remain active during the sensing phase.
772 \subsubsection{Coverage Ratio}
774 Figure~\ref{fig3} shows the average coverage ratio for 150 deployed nodes. It
775 can be seen that both DESK and GAF provide a little better coverage ratio
776 compared to DiLCO in the first thirty periods. This can be easily explained by
777 the number of active nodes: the optimization process of our protocol activates
778 less nodes than DESK or GAF, resulting in a slight decrease of the coverage
779 ratio. In case of DiLCO-2 (respectively DiLCO-4), the coverage ratio exhibits a
780 fast decrease with the number of periods and reaches zero value in period {\bf
781 X} (respectively {\bf Y}), whereas the other versions of DiLCO, DESK, and GAF
782 ensure a coverage ratio above 50\% for subsequent periods. We believe that the
783 results obtained with these two methods can be explained by a high consumption
784 of energy and we will check this assumption in the next subsection.
786 Concerning DiLCO-8, DiLCO-16, and DiLCO-32, these methods seem to be more
787 efficient than DESK and GAF, since they can provide the same level of coverage
788 (except in the first periods where DESK and GAF slightly outperform them) for a
789 greater number of periods. In fact, when our protocol is applied with a large
790 number of subregions (from 8 to 32~regions), it activates a restricted number of
791 nodes, and thus allow to extend the network lifetime.
796 \includegraphics[scale=0.45] {R/CR.pdf}
797 \caption{The Coverage Ratio}
801 %As shown in the figure ~\ref{fig3}, as the number of subregions increases, the coverage preservation for area of interest increases for a larger number of periods. Coverage ratio decreases when the number of periods increases due to dead nodes. Although some nodes are dead,
802 %thanks to DiLCO-8, DiLCO-16 and DiLCO-32 protocols, other nodes are preserved to ensure the coverage. Moreover, when we have a dense sensor network, it leads to maintain the coverage for a larger number of rounds. DiLCO-8, DiLCO-16 and DiLCO-32 protocols are
803 %slightly more efficient than other protocols, because they subdivides
804 %the area of interest into 8, 16 and 32~subregions if one of the subregions becomes disconnected, the coverage may be still ensured in the remaining subregions.%
806 \subsubsection{Energy Consumption}
808 % MICHEL - TO BE CONTINUED
810 Based on previous results in figure~\ref{fig3}, we keep DiLCO-16 and DiLCO-32
811 and we compare their performances in terms of energy consumption with the two
812 other approaches. We measure the energy consumed by the sensors during the
813 communication, listening, computation, active, and sleep modes for different
814 network densities. Figure~\ref{fig95} illustrates the energy consumption for
815 different network sizes.
816 % for $Lifetime95$ and $Lifetime50$.
817 We denote by $DiLCO-/50$ (respectively $DiLCO-/95$) as the amount of energy
818 consumed during which the network can satisfy an area coverage greater than
819 $50\%$ (repectively $95\%$) and we refer to the same definition for the two
823 \includegraphics[scale=0.45]{R/EC.pdf}
824 \caption{The Energy Consumption}
828 The results show that DiLCO-16/50, DiLCO-32/50, DiLCO-16/95 and DiLCO-32/95
829 protocols are the most competitive from the energy consumption point of
830 view. The other approaches have a high energy consumption due to activating a
831 larger number of redundant nodes.
834 %In fact, a distributed method on the subregions greatly reduces the number of communications and the time of listening so thanks to the partitioning of the initial network into several independent subnetworks.
835 %As shown in Figures~\ref{fig95} and ~\ref{fig50} , DiLCO-2 consumes more energy than the other versions of DiLCO, especially for large sizes of network. This is easy to understand since the bigger the number of sensors involved in the integer program, the larger the time computation to solve the optimization problem as well as the higher energy consumed during the communication.
837 \subsubsection{Execution Time}
838 We observe the impact of the network size and of the number of subregions on the computation time. We report the average execution times in seconds needed to solve the optimization problem for the different approaches and various numbers of sensors.
839 The original execution time is computed on a laptop DELL with intel Core i3 2370 M (2.4 GHz) processor (2 cores) and the MIPS (Million Instructions Per Second) rate equal to 35330. To be consistent with the use of a sensor node with Atmels AVR ATmega103L microcontroller (6 MHz) and a MIPS rate equal to 6 to run the optimization resolution, this time is multiplied by 2944.2 $\left( \frac{35330}{2} \times \frac{1}{6}\right)$ and reported on Figure~\ref{fig8}.
843 \includegraphics[scale=0.45]{R/T.pdf}
844 \caption{Execution Time (in seconds)}
849 Figure~\ref{fig8} shows that DiLCO-32 has very low execution times in comparison with other DiLCO versions, because the activity scheduling is tackled by a larger number of leaders and each leader solves an integer problem with a limited number of variables and constraints. Conversely, DiLCO-2 requires to solve an optimization problem with half of the network nodes and thus presents a high execution time. Nevertheless if we refer to figure~\ref{fig3}, we observe that DiLCO-32 is slightly less efficient than DilCO-16 to maintain as long as possible high coverage. Excessive subdivision of the area of interest prevents to ensure good coverage especially on the borders of the subregions.
851 %The DiLCO-32 has more suitable times in the same time it turn on redundent nodes more. We think that in distributed fashion the solving of the optimization problem in a subregion can be tackled by sensor nodes. Overall, to be able to deal with very large networks, a distributed method is clearly required.
854 \subsubsection{The Network Lifetime}
855 In figure~\ref{figLT95}, network lifetime is illustrated for different network sizes. The term $/50$ (respectively $/95$) next to the name of the method refers to the amount of time during which the network can satisfy an area coverage greater than $50\%$ ($Lifetime50$)(repectively $95\%$ ($Lifetime95$))
859 \includegraphics[scale=0.45]{R/LT.pdf}
860 \caption{The Network Lifetime}
865 As highlighted by figure~\ref{figLT95}, the network lifetime obviously
866 increases when the size of the network increases. For the same level of coverage, DiLCO outperforms DESK and GAF for the lifetime of the network. If we focus on level of coverage greater than $95\%$, The subdivision in $16$ subregions seems to be the most appropriate.
869 % with our DiLCO-16/50, DiLCO-32/50, DiLCO-16/95 and DiLCO-32/95 protocols
870 % that leads to the larger lifetime improvement in comparison with other approaches. By choosing the best
871 % suited nodes, for each round, to cover the area of interest and by
872 % letting the other ones sleep in order to be used later in next rounds. Comparison shows that our DiLCO-16/50, DiLCO-32/50, DiLCO-16/95 and DiLCO-32/95 protocols, which are used distributed optimization over the subregions, are the best one because it is robust to network disconnection during the network lifetime as well as it consume less energy in comparison with other approaches. It also means that distributing the protocol in each node and subdividing the sensing field into many subregions, which are managed
873 % independently and simultaneously, is the most relevant way to maximize the lifetime of a network.
878 \section{\uppercase{Conclusion and Future Works}}
879 \label{sec:Conclusion and Future Works}
880 In this paper, we have addressed the problem of the coverage and the lifetime
881 optimization in wireless sensor networks. This is a key issue as
882 sensor nodes have limited resources in terms of memory, energy and
883 computational power. To cope with this problem, the field of sensing
884 is divided into smaller subregions using the concept of divide-and-conquer method, and then a DiLCO protocol for optimizing the coverage and lifetime performances in each subregion.
885 The proposed protocol combines two efficient techniques: network
886 leader election and sensor activity scheduling, where the challenges
887 include how to select the most efficient leader in each subregion and
888 the best representative set of active nodes to ensure a high level of coverage.
889 We have compared this method with two other approaches using many metrics as coverage ratio, execution time, lifetime.
890 Some experiments have been performed to study the choice of the number of
891 subregions which subdivide the sensing field, considering different network
892 sizes. They show that as the number of subregions increases, so does the network
893 lifetime. Moreover, it makes the DiLCO protocol more robust against random
894 network disconnection due to node failures. However, too much subdivisions
895 reduces the advantage of the optimization. In fact, there is a balance between
896 the benefit from the optimization and the execution time needed to solve
897 it. Therefore, the subdivision in $16$ subregions seems to be the most appropriate.
899 \noindent In this paper, we have addressed the problem of the coverage and the lifetime
900 optimization in wireless sensor networks. This is a key issue as
901 sensor nodes have limited resources in terms of memory, energy and
902 computational power. To cope with this problem, the field of sensing
903 is divided into smaller subregions using the concept of divide-and-conquer method, and then a DiLCO protocol for optimizing the coverage and lifetime performances in each subregion.
904 The proposed protocol combines two efficient techniques: network
905 leader election and sensor activity scheduling, where the challenges
906 include how to select the most efficient leader in each subregion and
907 the best representative active nodes that will optimize the network lifetime
908 while taking the responsibility of covering the corresponding
909 subregion. The network lifetime in each subregion is divided into
910 rounds, each round consists of four phases: (i) Information Exchange,
911 (ii) Leader Election, (iii) an optimization-based Decision in order to
912 select the nodes remaining active for the last phase, and (iv)
913 Sensing. The simulations show the relevance of the proposed DiLCO
914 protocol in terms of lifetime, coverage ratio, active sensors ratio, energy consumption, execution time, and the number of stopped simulation runs due to network disconnection. Indeed, when
915 dealing with large and dense wireless sensor networks, a distributed
916 approach like the one we are proposed allows to reduce the difficulty of a
917 single global optimization problem by partitioning it in many smaller
918 problems, one per subregion, that can be solved more easily.
920 In future work, we plan to study and propose a coverage optimization protocol, which
921 computes all active sensor schedules in one time, using
922 optimization methods. \iffalse The round will still consist of 4 phases, but the
923 decision phase will compute the schedules for several sensing phases
924 which, aggregated together, define a kind of meta-sensing phase.
925 The computation of all cover sets in one time is far more
926 difficult, but will reduce the communication overhead. \fi
928 \section*{\uppercase{Acknowledgements}}
929 \noindent As a Ph.D. student, Ali Kadhum IDREES would like to gratefully acknowledge the University of Babylon - IRAQ for the financial support and Campus France for the received support.
936 \bibliographystyle{apalike}
938 \bibliography{Example}}
