Improving Lifetime of Large-scale Wireless Sensor Networks through Heterogeneity

Vivek Katiyar

Department of Computer Science and Engineering

National Institute of Technology Hamirpur (H.P.), INDIA

vivek.kat@gmail.com

Narottam Chand Department of Computer Science and

Engineering National Institute of Technology

Hamirpur (H.P.), INDIA nar.chand@gmail.com

Surender Soni Department of Computer Science and

Engineering National Institute of Technology

Hamirpur (H.P.), INDIA surender.soni@gmail.com

Abstract— In last few years, there has been a growing interest in the area of Wireless Sensor Networks (WSNs). Now a day’s WSNs are being deployed in large sensing areas for real life applications [1]. As sensor networks has limited and non-rechargeable energy resources, prolonging the lifetime of WSNs is a great issue. Some sensor nodes of comparatively high energy can be used (i.e. making the WSNs heterogeneous in terms of energy) to prolong the lifetime of WSNs. In this paper we propose an energy efficient protocol to prolong the lifetime of WSN by energy consumption through heterogeneity. The proposed approach selects cluster heads based on residual energy of sensor nodes. Simulation results demonstrate that heterogeneity in terms of energy of sensor nodes can improve lifetime of WSNs. Proposed protocol is also compared with LEACH, a previously existing clustering protocol for homogeneous WSNs.

Keywords- energy efficiency; heterogeneity; sensor node; WSN

I. INTRODUCTION In recent years, with the advances in the technology of

micro-electromechanical system (MEMS) and developments in wireless communications, wireless sensor networks (WSNs) have gained worldwide attention. WSNs consist of small nodes (sensors) having capabilities like sensing, computation, and communications. These sensors gather data by sensing the surroundings, aggregate this data to form useful information and transmit it to the base station or the neighboring node. WSNs have great potential for many application scenarios such as military sensing [2], industrial process monitoring and control [3], machine health monitoring [4], environment and habitat monitoring [5], healthcare applications [6], home automation and traffic control [7].

Large scale WSNs for real life applications can be formed with small sensor nodes scattered across some environment, as the advances in MEMS, VLSI technology and wireless communication technology. Sensor nodes in WSNs are limited in power, memory and computational capacity. So they may be short lived. A smart way to prolong the lifetime of sensor nodes and WSN as well is to make efficient use of energy. Many energy efficient algorithms have been proposed. LEACH [8] is very popular one of such algorithms that balance the energy consumption by rotating the cluster head (CH) role to every node in the cluster.

For efficient communication, these sensor nodes need to be in touch with far-away base stations. It can be achieved through clustering in WSN. Clustering is an important technique to prolong the lifetime of WSNs and to reduce energy consumption as well, by topology management and routing. Many energy-efficient routing protocols are designed based on the clustering structure [9, 10]. The clustering technique can also be used to perform data aggregation [8, 11]. Data aggregation is to combines the data from source nodes into a small set of meaningful information hence the fewer messages are transmitted thus saving communication energy. Within the clusters, localized algorithms can also efficiently operate. Clustering technique can be extremely effective in broadcast and data query [12, 13]. Cluster-heads help to broadcast messages and collect interested data within their own clusters.

Another way to prolong the lifetime of WSN is to insert a percentage of sensor nodes equipped with additional energy resources i.e. making the WSN heterogeneous in terms of energy. Even though the nodes may have same initial energy, energy of each node will not be same due to radio communication characteristics. Therefore, WSNs are more possibly heterogeneous networks rather than homogeneous ones. Many existing schemes for heterogeneous wireless sensor networks (HWSNs) like SEP [14], EEHC [15], DEEC [16], DEBC [17] etc., demonstrate that HWSNs survive for a longer time as compared to homogeneous WSNs.

In this paper, we propose an approach that prolongs the lifetime of wireless sensor networks through heterogeneity. The proposed clustering scheme forms clusters in each round by rotating the CH role among all nodes. CHs are elected depending upon the probability based on the remaining energy of the nodes. Simulation results show that proposed scheme can prolong the network life time and stability period. Results are also compared with LEACH.

Rest of the paper is organized as follows. Related work is described in Section II. Section III presents the energy model for heterogeneous WSN. Our proposed algorithm is described in section IV. Section V presents the simulation environment and results. We conclude the paper in section VI.

PROCEEDINGS OF ICETECT 2011

978-1-4244-7926-9/11/$26.00 ©2011 IEEE 1032

II. RELATED WORK Researchers generally assume that the nodes in wireless

sensor networks are homogeneous, but in reality, homogeneous sensor networks hardly exist. Even homogeneous sensors have different capabilities like different levels of initial energy, depletion rate, etc. The impact of heterogeneity of nodes (in terms of energy) on the performance of networks (energy-efficiency and stability) of various sizes and designs is evaluated in [18]. It is also known that if in a network, heterogeneity is used properly, then the response of the network is tripled and the network’s lifetime can be increased by 5-fold [18].

Smaragdakis G. et al. [14] proposed Stable Election Protocol (SEP), a heterogeneity-aware protocol. It is based on assigning weighted election probabilities of each node to be elected as cluster head. The cluster head is randomly selected based on the fraction of energy of each nodes, this assures that each node’s energy is uniformly used. In SEP, two types of nodes (normal and advanced) are considered. It is based on weighted election probabilities of each node to become cluster head according to the remaining energy in each node. This prolongs the stability period i.e. the time interval before the death of the first node.

One major drawback of this protocol is that sometimes no CH is selected. The data packets cannot be transmitted to the base station. This is a great disadvantage to the reliable transmission in the networks, especially for some important real-time transmission tasks.

Dilip and Patel [15] proposed an energy efficient clustered scheme for HWSNs based on weighted election probabilities of each node to become CH. It elects the CH in distributed fashion in hierarchal WSN. This algorithm is based on LEACH (Low Energy Adaptive Clustering Hierarchy) [8]. This algorithm works on the election processes of the CH in presence of heterogeneity of nodes.

A probability based clustering algorithm for HWSN named Distributed Energy-Balance Clustering algorithm (DEBC) [17], is proposed by C. Duan and H. Fan. DEBC elects cluster heads based on the knowledge of the ratio between remaining energy of node and the average energy of the network. This protocol also considers two-level heterogeneity and then it extends the results for multi-level heterogeneity. DEBC is different from LEACH, which make sure each node can be cluster head in each ni=1/p rounds.

Qing, Zhu and Wang [16] proposed DEEC; a distributed multilevel clustering algorithm for heterogeneous wireless sensor networks. DEEC selects the cluster heads with the help of probability based on the ratio between residual energy of each node and the average energy of the network. How long different nodes would be cluster heads, is decided according to the initial and residual energy. DEEC is also based on LEACH; it rotates the cluster head role among all nodes to expend energy uniformity. Two levels of heterogeneous nodes

are considered in this algorithm and after that a general solution for multilevel heterogeneity is obtained.

In DEEC, all the nodes need to know the total energy and lifetime of the network. When a new epoch begins, each node si computes the average probability pi by the total energy Etotal, while estimated value R of lifetime is broadcasted by the base station. Now pi is used to get the election threshold T(si). This threshold decides node si to be a cluster-head in the current round.

III. ENERGY MODEL FOR HETEROGENEOUS WSNS The model proposed here for heterogeneous WSN is same

as [16] that considers two types of nodes: normal nodes and advance nodes. As shown in Figure 1, normal nodes and advance nodes are represented by circle and star respectively.

Figure 1. A typical heterogeneous wireless sensor network with normal nodes and advance nodes.

It is assumed that out of total N nodes in the network, there are N∗− )1( θ normal nodes and N∗θ advance nodes. Advance nodes have γ times more energy than the normal nodes. The initial energies of normal nodes and advance nodes are iniE and iniE∗γ respectively. Hence total energy of network

iniinitotal ENENE ∗∗∗+∗−∗= γθθ )1( )1( γθθ ∗+−∗= iniEN

)1( −∗∗∗+∗= γθiniini ENEN This shows that heterogeneous network energy has

)1( −∗ γθ more virtual nodes than homogeneous networks with the energy iniEN ∗ .

The same energy dissipation model as in [8] is used here. To achieve an acceptable signal-to-noise ratio (SNR) in transmitting k bit message over a distance d, the energy cost of transmission )( TxE and reception )( RxE are given by:

1033

⎪⎩

⎪⎨⎧

∗∗+∗

∗∗+∗=

4

2

),(dkEk

dkEkdkE

mpelect

fselectTx ε

ε

Where nJE elect 50= , is the energy being used to run transmitter and receiver circuit. 2//10 mbitpJfs =ε and

4//0013.0 mbitpJmp =ε are the energies used by transmission amplifiers for short distance and long distance respectively. For

0dd = we can calculate the distance threshold,

mpfsd εε=0. To receive a k bit message, the

radio expends electRx EkE ∗= . FE is the energy of fusion per bit.

IV. PROPOSED ALGORITHM In this section we propose a clustering algorithm for WSN

that improve the lifetime through heterogeneity. Most of the proposed clustering schemes consider WSN to be homogeneous. Our proposed algorithm deals with the adaptation of the CH election process considering heterogeneous nodes in a clustering scheme.

Generally large scale WSNs are deployed for real life applications. The following assumptions are made about the sensor nodes and the network model:

1. The base station (i.e. sink node) is located inside the sensing field.

2. Nodes are deployed randomly in a square area. 3. Communication within the square area is not subjected to

multipath fading. 4. Nodes in the sensor field are heterogeneous in terms of

energy. 5. Two type of nodes: normal nodes and advance nodes. 6. The communication channel is symmetric. 7. Data gathered can be aggregated into single packet by

CH. Our proposed algorithm uses the idea of TRMRP of

PAMC, described in [19]. PAMC uses the concept of Minimum Reachability Power (MRP). MRP can be considered as transmission range. Distance parameter ∑ ),(1 skv nndis of EEMC, is replaced by ∑ vP1 , the Total Reciprocal of Minimum Reachability Power (TRMRP) for all active nodes. ∑= iMinPowTRMRP 1 , where MinPowi is minimum power level required by the node i to reach its CH. PAMC protocol also assumes that for any power level Li, there is corresponding transmission range Ri such that

Ri < Rj ∀ Li < Lj

The operation of the proposed algorithm can be divided into following two phases:

(i) Cluster set-up phase Initially, for each round, all nodes are set to regular (non-

CH). Cluster set-up phase is initiated by base station by sending a Start beacon. All nodes select its minimum power

level needed to reach the base station. The process of MRP discovery to base station is done only once during network life time and cached for subsequent usage. As active nodes receive the Start message, they reply by sending their residual energy

)(tEu and the MRP uP to the sink node to indicate that the CH will now be selected. On receipt of these values, sink broadcasts a ‘command’ message with values like total remaining energy of the network )(tE v and TRMRP ∑ vP1 . With help of these values each active node u sets its probability of becoming CH. Since network is heterogeneous, nodes having more energy have greater chances to become CH. The probabilities of becoming CH for normal nodes

)(uP n and advance nodes )(uP a can be given by following equations,

⎥⎦

⎤⎢⎣

⎡⎟⎠⎞

⎜⎝⎛∑

−+∑=∈∈ )()( /1

/1)1()( )()(

tSvtSv v

u

vu

tEtEopt

CHn

PPNuP φφ

⎥⎦

⎤⎢⎣

⎡⎟⎠⎞

⎜⎝⎛∑

−+∑=∈∈

∗

)()( /1/1)1()( )(

)(

tSvtSv v

u

vu

tEtEopt

CHa

PPNuP φγφ

Here, φ is a parameter deciding the weightage of energy factor and distance factor, and opt

CHN is the optimal number of CHs. Here ∑ ∈

∗+−∗∗=)(

)1()()(tSv uv tENtE γθθ and

)(tS is assumed to be the active node set of the network at time t. All CHs send out a CH message associated with MRP. Any node that receives this message replies with a “join” message sending out its residual energy and MRP to the CH.

(ii) Network operation phase When clustering topology is formed, the regular nodes start

transmitting the sensed data to their CHs. CHs aggregate the sensed data and send it to sink node. The cost of delivering data to the sink is sum of energy spent by nodes to send the data to their respective CH. It may be noted that the process of storing some of the last known MRPs can greatly reduce the communication cost of the network.

V. SIMULATION AND RESULTS In this section we evaluate the performance of our

proposed protocol. To validate the performance we have done simulation using ns-2 [20], a discrete event network simulator. Advance and normal nodes are randomly distributed over the field. We have compared the performance of proposed scheme with LEACH. The basic parameters used are listed in Table I.

We simulate proposed algorithm and Low-Energy Adaptive Clustering Hierarchy (LEACH) clustering algorithm for HWSN. Results from many runs of each algorithm are recorded for random distribution of nodes.

if d ≤ d0

if d ≥ d0

1034

TABLE I. SIMULATION PARAMETERS

Parameter Value Number of nodes 200-500 Network grid 100 X 100 m2 Base station position 50 X 50 m

fsε 10 pJ/bit/m2

FE 5 nJ/bit

electE 50 nJ/bit

Size of data packet 500 bits Initial energy of normal nodes 0.1 J Transmission range of sensor node 10 m

A. Network Lifetime We used the number of alive nodes parameter to evaluate

the network lifetime. For heterogeneous environment we have two types of nodes: advance nodes and super nodes. We carried out the simulation for 300 nodes with θ=0.2 and γ=3. It is clear from Figure 2 that our proposed algorithm outperforms LEACH in terms of network lifetime. This shows that the number of alive nodes is less after each round in LEACH. In presence of heterogeneity there are more chances for an advance node to become CH. So it is obvious that the time for nodes to die will increase.

0 100 200 300 400 5000

50

100

150

200

250

300

350

Num

ber o

f aliv

e no

des

Number of rounds

Proposed LEACH

Figure 2. Number of alive nodes per round.

B. Efficiency We evaluate and compare the efficiency of our proposed

algorithm in terms of average energy utilization ratio. It is calculated as the ratio of the total current consumed energy of all nodes to the total energy of all nodes at start of the simulation. Less the average energy utilization ratio, better the algorithm.

Figure 3 shows the average energy utilization ratio comparison. Due to heterogeneous nodes, network has comparatively high energy the heterogeneous networks survive for longer times. Proposed algorithm dissipates network energy in a slow rate because of heterogeneous nodes, hence performs better in comparison to LEACH in presence of heterogeneity.

We performed simulation for 200 nodes. Energy utilization ratio equal to zero means that no energy is expended till now and energy utilization ratio equal to one means no energy remaining i.e. sensor network is dead.

0 100 200 300 400 500 600 700 800 900 10000.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

Ave

rage

Ene

rgy

Util

izat

ion

Rat

io

Number of Rounds

Proposed LEACH

Figure 3. Average energy utilization ratio per round.

C. Number of Clusters per round We perform the simulation for 100 nodes. If we look at

Figure 4, number of clusters per round increase for proposed scheme. It takes the full advantage of heterogeneity (extra energy of advance nodes), the stable region is increased in comparison to LEACH. Proposed algorithm outperforms because advance nodes follow the death process of normal nodes, as the probability of electing CHs causes the energy of each node to be consumed in proportion to the nodes residual energy.

0 100 200 300 400 5000

5

10

15

20

25

30

35

Num

ber o

f clu

ster

s

Rounds

Proposed LEACH

Figure 4. Number of clusters per round.

VI. CONCLUSION The wireless sensor networks have been designed to help

in various monitoring applications. For real life deployments of large scale WSNs, network lifetime is big issue. We

1035

proposed to introduce some nodes of greater capacity in the sensor networks that results into an overall improvement in lifetime of WSNs. The proposed scheme uses the idea of minimum reachability power (MRP) for CH selection in heterogeneous environment. Simulation results show that network will survive for a long time in presence of heterogeneity capered to homogeneous WSNs.

REFERENCES [1] D. Culler, D. Estrin, and Srivastava, “Overview of Sensor Networks,”

IEEE Computer, Special Issue in Sensor Networks, August 2004, pp. 41-49.

[2] Chien-Chung Shen, C. Srisathapornphat and C. Jaikaeo, “Sensor Information Networking Architecture and Applications,” IEEE Personal Communications, August 2001, pp. 52-59.

[3] R. Kay and F. Mattern, "The Design Space of Wireless Sensor Networks", IEEE Wireless Communications, Vol. 11, No. 6, December 2004, pp. 54–61.

[4] S. Hadim, and N. Mohamed, "Middleware: middleware challenges and approaches for wireless sensor networks," Distributed Systems Online, IEEE, Vol.7, No.3, March 2006, pp.1-1.

[5] J.K. Hart and K. Martinez, “Environmental Sensor Networks: A revolution in the earth system science,” Earth-Science Reviews, 78, 2006, pp. 177-191.

[6] H. Yan, Y. Xu, M. Gidlund, "Experimental e-Health Applications in Wireless Sensor Networks," International Conference on Communications and Mobile Computing, 2009, pp. 563-567.

[7] J. Chinrungrueng, U. Sununtachaikul and S. Triamlumlerd, "A Vehicular Monitoring System with Power-Efficient Wireless Sensor Networks," ITS Telecommunications Proceedings, 6th International Conference on, June 2006, pp.951-954.

[8] W.R. Heinzelman, A.P. Chandrakasan, H. Balakrishnan, Energyefficient communication protocol for wireless microsensor networks, in: Proceedings of the 33rd Hawaii International Conference on System Sciences (HICSS-33), January 2000.

[9] P. Krishna, N.H. Vaidya, M. Chatterjee, D. Pradhan, A cluster-based approach for routing in dynamic networks, ACM SIGCOMM Computer Communication Review, Vol. 27, No. 2, 1997, pp. 49–65.

[10] B. McDonald, T. Znati, Design and performance of a distributed dynamic clustering algorithm for Ad-Hoc networks, in: Proceedings of the Annual Simulation Symposium, 2001.

[11] V. Mhatre, C. Rosenberg, D. Kofman, R. Mazumdar, N. Shroff, Design of surveillance sensor grids with a lifetime constraint, in: 1st European Workshop on Wireless Sensor Networks (EWSN), Berlin, January 2004.

[12] Sze-yao Ni, Yu-chee Tseng, Yuh-shyan Chen, Jang-ping Sheu, The broadcast storm problem in a mobile Ad Hoc network, in: Proceedings of the Annual ACM/IEEE International Conference on Mobile Computing and Networking (MobiCom’99), August 1999, pp. 151–162.

[13] D. Estrin, R. Govindan, J. Heidemann, S. Kumar, Next century challenges: scalable coordination in sensor networks, in: Proceedings of the 5th Annual ACM/IEEE International Conference on Mobile Computing and Networking (MobiCom’99), August 1999, pp. 263–270.

[14] G. Smaragdakis, I. Matta and A. Bestavros. “SEP: A stable election protocol for clustered heterogeneous wireless sensor networks,” In: Proc. of the International Workshop on SANPA, 2004, pp. 251-261.

[15] D. Kumar, T. C. Aseri and R.B. Patel, "EEHC: Energy efficient heterogeneous clustered scheme for wireless sensor networks", Computer Communications, Vol. 32, No. 4, 2009, pp. 662-667.

[16] L. Qing , Q. Zhu and M. Wang, "Design of a distributed energy-efficient clustering algorithm for heterogeneous wireless sensor networks", Computer Communications, Vol. 29, No. 12, August 2006, pp. 2230-2237.

[17] C. Duan and H. Fan, "A Distributed Energy Balance Clustering Protocol for Heterogeneous Wireless Sensor Networks," International Conference on Wireless Communications, Networking and Mobile Computing, 2007. WiCom 2007, Sept. 2007, pp. 2469-2473.

[18] M. Yarvis, N. Kushalnagar and H. Singh, “Exploiting heterogeneity in sensor networks,” IEEE INFOCOM, 2005.

[19] S. Soni and N. Chand, “ Energy Efficient Multilevel Clustering to Prolong the Lifetime of Wireless Sensor Networks,” Journal of Computing, Vol. 2, No. 5, May 2010, pp. 158-165.

[20] VINT Project. The ucb/lbnl/vint network simulator-ns. http://www.isi.edu/nsnam/ns.

1036