AbstractâIn a Wireless Sensor Network (WSN) hundreds of tiny sensors with .... they can have more energy for useful task of data transfer. But cluster head ...
2013 IEEE 9th International Conference on Mobile Ad-hoc and Sensor Networks
Energy Efficient Clustered Routing for Wireless Sensor Network Meenakshi Tripathi,R.B.Battula,M.S.Gaur,V.Laxmi Department of Computer Engineering Malaviya National Institute of Technology Jaipur,India nodes. In LEACH-C [5] the cluster head selection is done by the central authority i.e. base station and then the information about various cluster heads and their members are broad-casted to the network. The base station obtains the residual energy and x-y coordinates of all the nodes periodically and then selects only specific nodes having an energy value above average as eligible cluster heads.Now it applies simulated annealing algorithm to find the actual cluster head, which generate an optimal solution out of various random perturbations. In this process, the network may not perform optimally unless for a node the probability of selecting as a cluster head is calculated intelligently. The main objective of proposed work is to modify LEACH-C algorithm to achieve energy efficiency. The remaining paper is organized like this: Section IIbriefly describe the related work already present in literature.Section IIIpresents the system model and basic preliminaries of proposed work. A modified algorithm for selection of cluster heads has also been presented in the same section. Architecture and implementation details are discussed in Section IV.Performance evaluation is done in Section Vand finally Section VIconcludes the paper with future directions.
Abstract—In a Wireless Sensor Network (WSN) hundreds of tiny sensors with limited resources are accommodated to sense the information from the field. Transfer of gathered information from the sensing field to the base station must be done in proficiently to sustain the network longer. Clustering of sensor nodes is one way to achieve this goal. This paper introduces an Energy Efficient clustered routing protocol based on LEACH-C for WSN. In LEACH-C (Low Energy Adaptive Clustering HierarchyCentralized), the cluster heads are selected by the base station randomly. This paper introduces a novel cluster based routing protocol in which, the base station finds the highest energy node among the cluster and mark it as a cluster head for the current time. Thus in the proposed system the energy consumption of various nodes becomes more uniform as compared to LEACHC. The simulation results indicate that our proposed method leads to efficient transmission of data packets with less energy and therefore increases the network longevity as compared to LEACH-C and LEACH. Keywords—WSN, Cluster, LEACH-C, Energy Efficiency, NS2
I.
I NTRODUCTION
In a Wireless Sensor Network (WSN) numbers of battery operated sensors are distributed in the sensing region to sense changes in environmental or physical conditions like pressure, motion, humidity, temperature etc. These sensor nodes have limited resource like power, processing capacity and storage.The batteries of these nodes cannot be recharged or replaced and most of the time they are deployed in unattended areas, hence, efficient energy routing is must in increasing the lifetime of this kind of network. Many researchers have found that the hierarchical routing and specifically the clustered based routing play an important role in reducing energy depletion and increasing the network endurance ( [1], [2], [3]). Clustering is the processes of grouping the nodes based on some specifications and then find out the leader of every group known as cluster head. Cluster head has a responsibility to aggregate the data received by the nodes of respective group and send it to the base station (BS). In the literature of WSN several cluster based protocols are proposed like (LEACH [4],LEACH-C [5], HEED [6], PEGASIS [7]) etc. Its various applications like target tracking [8], environmental monitoring [9] and habitat monitoring [10] require only the aggregated value to be reported at the base station.This extra responsibility leads to early death of cluster head due to high energy dissipation.One of the most accepted cluster based routing protocol LEACH, rotates the responsibility of cluster head randomly among all sensor nodes to minimize this effect.Although in LEACH selection of cluster head is done in a distributed way, but it consumes lots of energy of sensor 978-0-7695-5159-3/13 $31.00 © 2013 IEEE DOI 10.1109/MSN.2013.67
II.
R ELATED W ORK
Direct transmission between the source node and the base station is the easiest way to deliver a packet between them. But,if the base station and the source node are far apart, then the energy consumption for transmitting the packets will increase proportionately. This direct transmission will also cause congestion at the base station ultimately resulting in potential data loss.The main task of the clustering algorithm is to select a leader node known as CH, which collect, aggregate and send the data from its member nodes to the base station. This reduces the congestion as well increases the lifespan of the nodes( [6], [7], [11], [12]). Some of these clustered routing protocols are discussed in the here. In LEACH [4] the cluster heads are chosen randomly.The selection of cluster head does not involve any central control. The cluster head selection algorithm of LEACH is a distributed algorithm as each node decides on its own to act as a cluster head for the particular time period. The node elects himself on behalf of predetermined number of cluster heads and its previous history as a cluster head. The cluster heads are changed after every round to preserve the battery power. It works in two stages: •
330
Set-up stages:The cluster heads and their members are decided than the cluster head sends the TDMA
cluster head is triggered after every Tround seconds. Each node joins the nearest cluster head so that it can reduce the transmission cost. Only the cluster heads can directly communicate with base station and hence, consumes a large amount of energy. In LEACH-C, the candidate cluster head must have energy value above the average energy of the network. Now the simulated annealing algorithm decides the actual cluster heads and their members for the next round. It may happen that many nodes with higher energy can not be selected as cluster head which causes the earlier death of less energy nodes in some rounds. The proposed clustering protocol LEACH-CE is aimed to choose the cluster head with higher residual energy in all the rounds to make it an energyefficient protocol, avoiding the fast energy depletion of sensor nodes. We name the protocol LEACH-CE , from the initials of the words LEACH-Centralized Efficient LEACH-C protocol. LEACH-CE exploits the residual energy of nodes to achieve network lifetime prolongation. One of the main parts of the proposed protocol is the selection of cluster heads.
schedule to their cluster members so that they can transmit the data in allotted slot. •
Steady-state stage: Cluster members transmit data in allotted slot to respective cluster heads, where they perform the aggregation and transmit the aggregated data to the base station.
While selecting the cluster heads LEACH does not take into account the residual energy of a node, which may cause early death of cluster head having little residual energy, thus shorten the entire network lifetime. In [13] the authors have proposed a Multi-hop Routing with LEACH known as MR-LEACH.Whole network is divided into several layers and every layer has a cluster of nodes. Cluster head of each layer transmit the data to the head of adjacent layers to until it reaches to the base station. On the basis Received Signal Strength other nodes join the cluster head. Base Station controls the complete transmission and distribute Time Division Multiple Access (TDMA) schedule for each cluster-head. Cluster heads of upper layer is selected as super cluster heads for lower layer cluster heads by the base station. Since the data has to go through the multiple layers before reaching to the base station so it increases the probability of message failures. LEACH-C [5] is an improved version of LEACH, in which the cluster formation is done by the base station. All the nodes send their geographical coordinates and residual energy to the base station at the start of a new round. The base station determines the average energy of the network and marks only those nodes which are having energy higher than the average, as eligible cluster head node. Now it applies simulated annealing [14] algorithm using candidate nodes to minimize the objective function. This algorithm minimizes the total sum of squared distances all the non-cluster head nodes and the closest cluster head and hence, reduces the energy consumption. The resultant cluster head (CH) and their members will be broadcasted to the network. If the node’s own ID matches with cluster head ID, it elects himself as cluster head otherwise it will find out the TDMA slot to transmit the data to corresponding cluster head. The data transmission phase of LEACH-C is similar to the LEACH. The advantage of LEACH-C is that, it does not put the burden of cluster formation on resource constraint sensor nodes so they can have more energy for useful task of data transfer. But cluster head selection does not guarantee the balance of energy consumption of whole sensor network. In [15] the authors have proposed a mechanism to forward the data the cluster heads of LEACH-C. A table is maintained by every cluster head which contains the routing information and distance from other cluster heads (CHi =1,2,3..........N). The cluster head x which is situated nearest to base station takes the responsibility of retransmission of all the data gathered by another cluster head y, which is the neighbor of x. This process helps in conserving the energy of cluster head y. But this protocol creates a burden of maintaining the routing table at each node, thus reduces the lifetime of the network. III.
A. System Model For network configuration we assume that WSN includes large numbers of sensor nodes, dispersed in a sensor field. The total number of sensor nodes in the field is equal to N. We did not make any assumptions about the network diameter and network density. Also, we consider the sensor network posses the following properties: •
After initial deployment neither sensor nodes nor base station has any movement. In the most of the applications, sensor nodes have no mobility. Initially the battery level of all sensor nodes is same.
•
Links are bidirectional.
•
The computation and communication capabilities of all the sensor nodes are same. Moreover, it is not feasible to recharge nodes batteries. For example in a battlefield, sensor nodes are dispersed in a large target area where reaching and recharging them is extremely difficult and dangerous. This prompts us to propose a protocol that is energy aware so that network can operate for longer time.
•
Nodes are location aware i.e. thus either equipped with a GPS device or use some method to find the location.
•
The nodes are considered to die only when their energy is exhausted.
•
All the nodes have data to send.
B. Performance Parameters Here we define the parameters used in this paper to exploit the effectiveness of proposed clustered protocol: •
Network endurance: The time difference of the commencement of sensor network and the death of the last active node.
•
Number of alive nodes: This measure indicates the total number of nodes that has not yet expended all of their energy.
T HE P ROPOSED M ETHOD
Hierarchical structure of our clustered Wireless Sensor Network is same as considered in LEACH or LEACH-C. In both of the clustering algorithms a process to select a new
331
•
FND (First Node Death Time): This is calculated by the difference of time from start of operation to the death of first node in the network.
•
Energy Consumed: This measure represents the total amount of energy expended by all the nodes of the network.
the middle of the arena, and that the distance of a sensor node to the sink or its cluster head is < d0 . Cluster head is receiving the data from its members, aggregating it and then sending to the base station. Thus, energy dissipation by a cluster head for one round is calculated by the formula given below: Ech = ERx ∗ (N/K − 1) + EAGGR + ET x (l, dBS )
C. Algorithm for Cluster Head Election We have used following assumptions for our calculations. We assume that in steady state phase, all member nodes transmit their data to the respective cluster head and after performing data aggregation or compression the cluster head nodes will transmit data to the base station. For the purpose of this research we use the first order RF model to estimate the energy consumption of various nodes as proposed in [16].It is assumed that nodes are not consuming any energy when not receiving or sending the data. Assume the energy dissipated by electric circuit to send or receive 1-bit data is Eelec , sender and receiver nodes are d distance apart from each other then the first order RF model of energy consumption is given by Figure 1.Where μamp is the energy consumed in transmitter amplifier to achieve acceptable bit energy to noise power spectral density ration at the receiver and it depend upon the model we use. μf s is for free-space model and μtr is for two-ray model.
Where K is expected number of cluster heads and EAGGR is the aggregation cost of l-bit data by the cluster head, which is given by: EAGGR = N/K ∗ l ∗ EDA
Ecm = l ∗ Eelec + l ∗ f s ∗ d2ch
If d ≤ d0 If d ≥ d0
•
Select initial cluster heads The initial K cluster heads will be selected randomly from the set of nodes whose residual energy is above the average energy of the network i.e. Ei > EAvg where N i=1 Ei (8) EAvg = N Ei is the residual energy of i’th node.
•
Assign the cluster members In order to create clusters the distance between the K cluster heads and all the sensor nodes is calculated. Then, the clusters are formed based on the minimum distance. If cluster head Cj is closest to the node Qi in t’th round then the conditions given below must be true for all i*=1,2,3...K.
(1)
(2)
by equating above equations at d=d0 we get d0 =( f s /tr ) The energy cosumed in receiving l-bit data is: ERx (l, d) = Eelec (l)
Set-up Phase: The main intent of this phase to group the nodes into various clusters and find their cluster heads. In this phase base station receives the location and residual energy information of all the sensor nodes in the network. If there are N nodes in the network distributed in an area M then number of clusters K is given by the following formula, which is defined in LEACH-C:√ N f s M K= √ (7) 2π tr d2toBS Where dtoBS is the average distance between cluster heads to the base station.
To attain tolerable Signal-to-Noise Ratio(SNR) in transmission of l-bit packet over distance d, the energy expended is given by:-
l ∗ Eelec + l ∗ f s ∗ d2BS ET x (l, d) = l ∗ Eelec + l ∗ tr ∗ d4BS
(6)
From above analysis we can say Ech = α ∗ Ecm where, α is a constant. The node that is becoming the cluster head must have highest residual energy to perform the task of aggregation and transmission. The major steps involved in our algorithm are described below:
Radio energy dissipation model
ET x (l, d) = Eelec (l) + amp (l, d)
(5)
EDA is the energy consumed by cluster head to fuse 1-bit data. Energy consumed by a cluster member in sending the l-bit data is given by:
•
Fig. 1.
(4)
(3)
Assume an area A = M × M square meters over which N nodes are evenly dispersed. We assume the sink is situated in
Cj (Xj ) = (Xj − Qi(X) )2 ≤ (Xj − Qi(X)∗ )2
332
(9)
Cj (Yj ) = (Yj − Qi(Y ) )2 ≤ (Yj − Qi(Y )∗ )2
D. ARCHITECTURE OF LEACH-CE
(10)
In primitive version of LEACH-C The eligibility of becoming a cluster head is not decided by the residual energy difference of various nodes, and as a result energy resources of the overall network is not well utilized. The reason is that LEACH-C is looks at average energy of the network and residual energy of the node. It does not compare the residual energy of the nodes with each other. So there is a chance that nodes having less energy become the cluster head for a round and dies easily. Let us assume that in our 100m × 100m sensor network the sensors are distributed randomly, as shown in Fig. 2.
So the node Qi will join cluster head Cj for the current round whose coordinates are : Cj = (Xj , Yj ) •
(11)
Cluster Head Selection Final cluster head selection from the already formed cluster is done on basis of residual energy of the all the nodes. Let Ejq = (E1 , E2 ............Ei ) is set of residual energy of all the cluster members of j’th cluster then the cluster head Cji will be the node having maximum energy among all the i nodes of that cluster.The following algorithm returns the ID of the node which has maximum energy in the cluster and act as a cluster head for the current round.CH is the set of cluster heads,CM is the set of members of various clusters and E is the set of energy of all the nodes. CH-Selection(CH,CM,E) 1: 2: 3: 4: 5: 6: 7: 8: 9: 10:
for l = 1 → k q←1 max for q = 2 → i if (Ejq+1 ≥ Ejq then max + 1 end if return max endfor endfor
Fig. 2.
Here,in detail our new energy efficient LACH-CE protocol is explained whose objective is to achieve longer network lifetime. Let us assume that initial energy of all the sensor nodes is same and they are distributed evenly over the sensing region. Our approach is to assign a priority pi to a node to become cluster head pch . This priority depends upon a factor Fi which is equal to the initial energy of a node divided by its residual energy. Only nodes which are having the highest priority will be eligible to become a cluster head in next round. The probability of a member node to die first is less than the probability of an already chosen cluster head node to die. Simulation results attest our expectation.
After finalization of cluster heads for every node base station broadcast this information in the network.A node checks the broad-casted information to find whether it needs to act as a cluster head or cluster member. Cluster head has to fix the time slots for its members to send the data in TDMA schedule.Algorithm 2 describe the proposed protocol in detail. Final-CH-Selection(E,N,K) 1: 2: 3: 4: 5: 6: 7: 8: 9: 10: 11: 12: 13: 14: 15: 16: 17:
Set-up Stage for i = 1 → N if (Ei ≥ EAvg then Eligible(i) ← T rue else Eligible(i) ← F alse end if endfor Select K CH randomly f or which Eligible(i) T rue for i = 1 → k (j+1) if (disti ≥ distji ) then CHi ← CMj ∀jN endfor end if for i = 1 → k CH ← CH − Selection(CH, CM, E) endfor
Sanpshot of Wireless Sensor Network
IV.
S IMULATION R ESULTS
The following section, evaluate the effect of the LEACHCE protocol on various network performance parameters. First order RF model is considered for the simulation of LEACH and the values of various simulation parameters are mentioned in Table 1. The performance of LEACH-CE is checked by simulating a wireless sensor network using NS-2 in a 100m × 100m sensing region with several scenarios. The total number of sensor nodes N = 100.
=
The sensor nodes are distributed in a random way over the entire region. That means the (x,y) coordinates of every sensor node lies between (0,0) and (100,100) randomly. The base station is in the middle of the field, so that maximum distance of a sensor from the base station is approximately 70 m. The message that is being transmitted by sensor nodes to their cluster heads and cluster head sends to base station of size 4000 bits. Simulation setup developed by us is same as used
333
Fig. 3.
Number of alive nodes with (a) initial energy 2J/node (b) Initial energy 1J/node
Fig. 4.
Total Energy Consumption (a) Initial energy 1J/node (b) Initial energy 2J/node
Description Number of nodes Amplifier energy dissipation to disseminate data at a short distance Amplifier energy dissipation to disseminate data at a long distance Energy consumed by electronic circuit to send or receive signal Energy consumption in data aggregation Crossover distance Message size
TABLE I.
Symbol N f s
Value 100 10pJ/b/m2
tr
.0013pJ/b/m4
Eelec
50nJ/b
Eelec d0
5nJ/b/message 70m 4000bits
Figure 4 shows that the total energy consumption of the network in LEACH-CE is lower than LEACH and LEACH-C in both the scenarios. It’s because in LEACH the distribution of cluster head is uneven, in the sense that, in some region the cluster heads are too concentrated, while some region do not have any cluster heads. Furthermore, some nodes are far away from the cluster heads which will conduce to the waste of energy. Although the situation has been improved in LEACH-C but LEACH-CE gives us better results compared with LEACHC. It is observed from Figure 5 that 1%,10%,20%,50% or 96% node death time increases in all the scenarios for the LEACHCE. So we can say that just by distributing the loads evenly among all the nodes LEACH-CE performs better than LEACH or LEACH-C in all the cases. For our simulation to stop we have two conditions either the simulation time gets over or the remaining alive nodes becomes less than K that is why we have taken the case of death of 96% nodes.
S IMULATION PARAMETERS
by authors of LEACH [4].We observed the effect on various network performance parameters when the initial energies of nodes were 1J and 2J. This section, examines the performance of network in presence of LEACH, LEACH-C and our proposed LEACH-CE protocol in the same simulation settings. It can be seen from the Figure 3 that the within the same time period number of nodes dead in LEACH and LEACH-C is more than LEACH-CE. The number of nodes dies very quickly and as a result the nodes get scattered in the field very fast. In LEACH-C random selection of cluster head is done from the set of candidate cluster heads so if same node becomes cluster head more than once then it dies very fast, on the other hand in LEACH-CE cluster head selection process is done on the basis of residual energy so the distribution of cluster head selection is more uniform. Hence, in LEACH- CE the nodes die at a slow rate. When the number of dead nodes reaches to a threshold (i.e. 96 for our scenario) network stops functioning.
V.
C ONCLUSIONS
Wireless sensor networks are used in lots of monitoring and control applications. Power saving in clustered routing is mandatory to extend the network lifetime. This paper contributes an energy efficient cluster based routing for Wireless Sensor Network. The lifetime of the network has been improved by selecting the cluster heads intelligently. While comparing the simulation results of LEACH-CE with LEACH, 10% improvement has been achieved, while with LEACH- C 5% improvement was there in terms of network lifetime. The
334
Fig. 5.
Performance results (a) Initial energy 1J/node (b) Initial energy 2J/node
reduction in total energy consumption of the whole network has also been obtained in the presence of same simulation settings. Hence we can say that proposed routing protocol performs better in terms of network lifetime and energy saving. We have compared our protocol with LEACH-C only, but it can also be compared with other clustered protocols.
[13]
[14]
R EFERENCES [15]
[1] Xuxun Liu. A survey on clustering routing protocols in wireless sensor networks. Sensors, 12(8):11113 –11153, 2012. [2] N. Vlajic and D. Xia. Wireless sensor networks: To cluster or not to cluster? In IEEE 14th International Symposium on A World of Wireless, Mobile and Multimedia Networks (WoWMoM), pages 258–268, Los Alamitos, CA, USA, June 2006. IEEE Computer Society. [3] C. Qing, T. Abdelzaher, H. Tian, and R. Kravets. Cluster-based forwarding for reliable end-to-end delivery in wireless sensor networks. In Annual Joint Conference of the IEEE Computer and Communications Societies INFOCOM, pages 1928–1936, Alaska, USA, May 2007. IEEE Computer Society. [4] W.B. Heinzelman, A.P. Chandrakasan, and H. Balakrishnan. Energy efficient communication protocol for wireless sensor networks. In Hawaii international conference on system sciences, pages 3005–3014, Island of Maui,Hawaii, January 2000. IEEE Computer Society. [5] W.B. Heinzelman, A.P. Chandrakasan, and H. Balakrishnan. An application-specific protocol architecture for wireless microsensor networks. IEEE Transactions on Wireless Communications, 1(4):660 –670, 2002. [6] O. Younis and S. Fahmy. Heed a hybrid, energy-efficient, distributed clustering approach for ad hoc sensor networks. IEEE Transactions on Mobile Computing, 3(4):366 –379, 2004. [7] Stephanie Lindsey and Cauligi S. Raghavendra. Pegasis power-efficient gathering in sensor information systems. In IEEE Aerospace Conference, Big Sky, MT, USA, March 2004. [8] P.K. Biswas and S. Phoha. Self-organizing sensor networks for integrated target surveillance. IEEE Transactions on Computers, 55(8):1033 –1047, 2006. [9] Francois Ingelrest, Guillermo Barrenetxea, Gunnar Schaefer, Martin Vetterli, Olivier Couach, and Marc Parlange. Sensorscope: Applicationspecific sensor network for environmental monitoring. ACM Transactions on Sensor Networks, 6(2):1033 –1047, February 2010. [10] R. Szewczyk, E. Osterweil, J. Polastre, M. Hamilton, A. Mainwaring, and D. Estrin. Habitat monitoring with sensor networks. Communications of the ACM, 47(6):34 –40, 2004. [11] L. Tao, QX Zhu, and L Zhang. An improvement for leach algorithm in wireless sensor network. In 5th IEEE Conference on Industrial Electrical Application, pages 1811–1815, Taichung, Taiwan, 2010. IEEE Computer Society. [12] Thiemo Voigt, Hartmut Ritter, Jochen Schiller, Adam Dunkels, and Juan Alonso. Solar–aware clustering in wireless sensor networks. In Ninth
[16]
335
IEEE Symposium on Computers and Communications, Alexandria, Egypt, June 2004. IEEE Computer Society. Muhamnmad Omer Farooq, Abdul Basit Dogar, and Ghalib Asadullah Shah. Mr–leach multi-hop routing with low energy adaptive clustering hierarchy. In fourth International Conference on Sensor Technologies and Applications, Venice, Italy, June 2010. IEEE Computer Society. T. Murata and H.Ishibuchi. Performance evaluation of genetic algorithms for flowshop scheduling problems. In 1st IEEE conference Evolutionary Computation, pages 812–817, Orlando, USA, June 1994. IEEE Computer Society. L. Layuan Y. Chun. A solution for the energy hole based on leach-c protocol, 2010. Javad Vazifehdan, R. Venkatesha Prasad, Martin Jacobsson, and Ignas G. Niemegeers. An analytical energy consumption model for packet transfer over wireless links. IEEE Communications Letters, 16(1):30 – 33, 2012.
