Transactions On Internet And Information Systems

Published Date: 02 Nov 2017

An event driven clustering routing protocol for wireless

sensor networks

Journal: KSII Transactions on Internet and Information Systems

Manuscript ID: TIIS-LT-2013-Feb-0139

Manuscript Type: Letter

Date Submitted by the Author: 24-Feb-2013

Complete List of Authors: Fu, Hongyu

Pan, Yijin

Li, Guoquan

Wu, Yucheng

Keywords of your Paper:

Wireless Sensor Networks, Routing Protocols and Congestion Control, Wireless Communications

http://mc.manuscriptcentral.com/tiisjournal

Page 1 of 13

KSII Transactions on Internet and Information Systems

An event driven clustering routing protocol for

wireless sensor networks

Fu Hongyu, Pan Yijin, Li Guoquan, Wu Yucheng

College of Communication Engineering, Chongqing University, Chongqing 400044, China

[e-mail:[email protected]; {panyijin, lgq, wuyucheng}@cqu.edu.cn]

*Corresponding author: Wu Yucheng

Abstract

Emergent event monitoring is one of the important applications of wireless sensor networks. This kind of application has more requirements in routing protocol due to the randomness on when and where the event happens. In this paper, a novel event driven clustering routing protocol (EDCR) is proposed for emergent event monitoring

applications of WSN, which is aimed at prolonging the effective working time of the

whole sensor network by holding over the lifespan of the first dead node. In EDCRï¼Œevent

driven clustering mechanism is applied to satisfy the randomness of emergent event and a new method is given to select the cluster head where the residual energy , the distance to the sink node and the connectivity with the neighbors are taken into consideration. In the data transmission phase, the current node selects the node with more residual energy and at a proper location as the next relay node according to its own amount of residual energy. The simulation results show that EDCR has a better performance on both energy economization and energy consumption balance compared to ARPEES and AEEC.

Keywords: wireless sensor network, event driven clustering, emergent event monitoring,

routing protocol.

http://mc.manuscriptcentral.com/tiisjournal

KSII Transactions on Internet and Information Systems Page 2 of 13

1ï¼ŽIntroduction

Self-organized and cooperation are two prominent features of wireless sensor networks

(WSN), which ensure a wide range of applications for WSN [1]. However, the limited

energy resource and processing capacity of sensor nodes decrease the performance of the

deployed sensor networks. Effective and applicable routing protocols can make up for the limitation of the single node, and improve the energy efficiency, and then prolong the effective working time of the whole network.

Hierarchical routing protocols are proven to be more energy-efficient than flat ones in

which all the nodes play the same role, especially in terms of the data aggregation and the

flooding of control packets [2]. Low Energy Adaptive Clustering Hierarchy (LEACH) [3]

is one of the most famous hierarchical routing protocols. In LEACH, all the sensors are

grouped into clusters while a few sensors are selected to be cluster heads (CHs), and other

normal sensors join in the cluster of the nearest CH. The CHs are responsible for collecting

sensed data from their own cluster members and aggregating the collected data, and then

transferring the aggregated data to the sink node. In order to balance energy consumption,

the CHs selection runs periodically and the node never being a CH is more probable to be

selected as a CH. There are still some disadvantages in LEACH though it improves

energy-efficiency to some extent. For instance, the CHs selection doesnâ€™t consider the

residual energy of sensors. All sensors send data during each frame. The CHs transmit data

packets to the sink node directly without considering the distance to the sink node. All of

these may result in early death of sensors. Recently, lots of protocols are proposed to

improve LEACH [4-8]. HEED [4] selects cluster heads periodically according to nodeâ€™s

residual energy and a secondary parameter, such as proximity to neighbors or degree. [5]

views sensing coverage as one of the critical measurements of performance offered by a

sensor network and proposes coverage-preserving routing protocols, which are modified

from the LEACH and virtual grid routing protocols. In [6], sensors are grouped into

several static clusters by the Hausdorff clustering algorithm based on node location,

communication efficiency, and network connectivity. Clusters are formed only once and

cluster heads are optimally scheduled among the cluster members periodically and form a

backbone network to collect, aggregate, and forward data to the base station using

minimum cost routing. AEEC [7] involves three main steps. In the initial phase, the sink

node selects the most separated k (optimal number) nodes as elector nodes. In the

clustering phase, the elector nodes take responsibility for collecting nearby sensorsâ€™ energy

information and selecting cluster head. The cluster heads collect sensed data from active

cluster members and then transmit the aggregated data to the sink node in the data

transmission phase. HEEP [8] makes a combination of LEACH and PEGASIS [9],

organizing the network nodes in chains clusters to avoid the bad energy dissipation in

LEACH and to reduce the routing delay generated by PEGASIS.

http://mc.manuscriptcentral.com/tiisjournal

Page 3 of 13

KSII Transactions on Internet and Information Systems

The protocols mentioned above have a common drawback in terms of emergent event monitoring applications that the formed clusters are not suitable for the randomness of where and when the events happen, which may reduce the efficiency of data collection and data aggregation, and lead to unnecessary energy consumption. [10] proposes an

event-driven clustering routing algorithm (EDC) which overcomes the common drawback

of the previous protocols. In EDC, the sensor broadcasts its remainder energy to a close

gateway node after it sensed some abnormal information. The gateway node takes

responsibility for collecting all related sensorsâ€™ energy information and sorting them. The

one with maximal remainder energy is selected as cluster head and the other related

sensors send their data to the cluster head. This protocol saves lots of energy because the

irrelevant sensors donâ€™t have to join the cluster. However the performance of this protocol

highly depends on the density and placement of the gateways, which may not be practical

for many WSN scenarios. ARPEES [11] makes all the sensors in sleep mode to save

battery power. When the event happens, the nodes near the event become activated, and

then an algorithm is executed to form a cluster and select a cluster head. In the data

transmission phase, multi-hop relay mechanism is applied. When selecting the relay nodes,

the trade-off relationship between the residual energy and the distance to the sink node are

considered. ARPEES shows a good performance on prolonging the lifetime of the whole

network. Nevertheless, the first dead node appears early after the network is deployed. In

[12], a novel event-to-sink directed clustering protocol (ESDC) is proposed which forms

clusters when and where needed and in the direction of data flow from event location to

the sink. [13] achieves continuous monitoring using an event-driven reporting approach

called CM-EDR. This mechanism is independent in the sense that it can be applied to any

reference protocol.

In this paper, we propose a novel event driven clustering routing protocol (EDCR) for emergent event monitoring applications of WSN, which is aimed at prolonging the effective working time of the whole sensor network by holding over the lifespan of the first dead node. EDRC applies event driven clustering mechanism to satisfy the randomness of emergent event and present a new method for selecting the cluster head which considers not only the residual energy of the node but also the distance to the sink node and the connectivity with neighbor nodes. In the data transmission phase, the current node selects the node with more residual energy and appropriate location as the next relay node according to its own amount of residual energy.

The remainder of the paper is organized as follows: Section 2 describes the model and some assumptions of the sensor network that we consider. The details of EDCR protocol is presented in Section 3. Performance analysis and simulation results are given in Section 4. Finally, the last section concludes this paper.

2. Network Model and Assumptions

2.1 Network Model and Assumptions

http://mc.manuscriptcentral.com/tiisjournal

KSII Transactions on Internet and Information Systems Page 4 of 13

In this paper, we have the following assumptions.

1) The only sink node locates at the center of the monitoring area and has enough energy,

memory and computing capability.

2) The sensor nodes are randomly deployed in the monitoring area. Equivalent

processing and communication capability and limited and equal initial energy are

assumed for each of them.

3) All nodes are immobile. Each node has a unique identifier (ID) and the ability to

adjust the transmission power according to the distance to the desired receiver.

2.2 Radio Model

The radio model of WSN consists of transmitter, power amplifier and receiver [3]. The energy consumption to transmit a k-bit packet over distance d is defined as

E ( k, d) = E ( k ) + E (k , d )

21 TX

TX âˆ’ elec TX âˆ’ amp

ï£´ E Ã— k + Îµ Ã— k Ã— d 2 , (

d < d ) (1)

23 = ï£²

24 ï£³

elec fs

E elec Ã— k + Îµ mp Ã— k Ã— d

, ( d â‰¥ d 0 )

27 where ETX-elec is the power required by the electronic devices to transmit one packet, which

28 is decided by the packet length k and the energy consumption per bit, denoted as Eelec.

30 ETX-amp is the amplifier energy consumption, and it depends on the distance d between the

31 transmitter and the receiver. When d<d0, a free space model is adopted and Îµfs is used to

33 calculate ETX-amp; When dâ‰¥d0, a multi-path model and Îµmp is the coefficient. d0 is a

35 threshold obtained from

38 d = Îµ / Îµ (2)

0 fs mp

The energy consumed for the receiver to process a k-bit packet is expressed as

ERX(k) = ERX-elec(k) = EelecÃ—k (3)

Define Eda as the energy needed to aggregate one bit, then the energy for data aggregation

EDA = EdaÃ—k (4)

2.3 Optimal Transmission Distance

Combing (1) and (3), an optimal transmission distance can be found to obtain the highest

http://mc.manuscriptcentral.com/tiisjournal

Page 5 of 13

KSII Transactions on Internet and Information Systems

energy-efficiency for unit distance by the following equation.

E / d = ( E + E ) / d

tf TX RX

ï£´ 2 Ã— E Ã— k / d + Îµ Ã— kÃ— d , ( d < d )

= ï£²

elec fs 0 (5)

ï£³ 2 Ã— E elec Ã— k / d + Îµ mp Ã— kÃ— d , ( d â‰¥ d 0 )

When we set

âˆ‚( E tf / d )

= 0 (6)

âˆ‚ d

And Eelec = 50nJ/bit, Îµfs = 10pJ/bit/m2, Îµmp = 0.0013pJ/bit/m4, we can get d=d0=87.7m. As

shown in Fig.1, there is a minimum of (Etf /d) when d = d0, which means that when the

distance between the transmitter and the receiver is d0, the consumed energy is lowest for

unit distance. Therefore, in our protocol, the transmission radiuses R0 of all the sensor

nodes are initialized as d0.

x 10-8

0 50 100 150 200 250 300

Transmitting distance (m)

Fig.1 The ratio of energy consumption and transferring distance with a single hop

3. Protocol Description

Since the first dead node will damage the connectivity of the neighbors, its lifecycle has a

great impact on the whole sensor network [14, 15]. In this paper, we proposed a new

routing protocol to prolong the lifecycle of the first dead node so that the whole sensor

networkâ€™s lifespan can be held over. In order to save the energy of the whole network,

http://mc.manuscriptcentral.com/tiisjournal

KSII Transactions on Internet and Information Systems Page 6 of 13

event-driven clustering mechanism [10] is applied, which can avoid unnecessary energy

consumption by preventing irrelevant sensor nodes taking part in the cluster formation.

Moreover, grouping all the sensors related to an event in one cluster is helpful to efficient

data compression [11]. In addition, both single-hop and multi-hop transmission are applied

depending on the distance between the transmitter and receiver, especially for the data

transmission between cluster heads and sink node. When we choose the relay node, not

only the condition of candidate nodes but also the residual energy of the current node will be considered. In this way, the sensor nodes with less residual energy can be saved from getting dead prematurely.

In the following, the main three parts of the proposed protocol, network initialization, cluster formation and data transmission, are discussed in details.

3.1 Network Initialization

After the network deployed in a monitoring area, the sink node broadcasts network initial

message (Net-Initial-M) with various power levels. All the sensor nodes calculate their

own distance to the sink node according to the power strength of the received

Net-Initial-M, and then all the sensor nodes broadcast node information message

(Node-information-M) with the default power level. The Node-information-M contains

some information of the node such as ID, residual energy and distance to the sink node.

However, there may be some isolated nodes under the condition of the default power level

and they should increase their own power levels properly to search for a nearby node

which is nearer to the sink node than itself as a neighbor node. Thus, every sensor node

can obtain the information of their neighbor nodes and know its distance to each neighbor.

3.2 Cluster Formation

The nearby sensors that sensed abnormal attribute values become related nodes when an event happens. Each related node calculates its probability of being the cluster head by (7), and the one with maximum Pch will be the cluster head of the current event.

E 1 N

P ch = k1 Ã—

res nb

+ k2 Ã— + k3 Ã— (7)

E d N

0 s

where Eres and E0 are the residual energy and initial energy of the node respectively. ds is the distance between the node and the sink node. Nnb is the number of neighbors of the current node. N is the total number of sensor nodes in the network, which is a default setting of each sensor. And ki (i=1,2,3) are the weights of the three factors respectively. Here we have ki>0 and Î£ki= 1, we can set ki according to the application scenarios.

The related nodes execute the following algorithm to select the cluster head and then form a cluster:

1) Each related node waits for a time segment Ti and Ti = (1-Pchi)Ã—T0, where i is the

related nodeâ€™s ID and T0 is a constant. It can be noticed that the one with lower

http://mc.manuscriptcentral.com/tiisjournal

Page 7 of 13

KSII Transactions on Internet and Information Systems

probabilities of becoming a cluster head will wait for longer time.

2) The related node broadcasts cluster head selection message (CH-Select-M) at the end

of Ti if it has not received any CH-Select-M from other related neighbors during Ti.

CH-Select-M contains the ID and Pch of the producer.

3) If the related node receives more than one CH-Select-M during Ti, records and

broadcasts the first one.

4) After a time segment of TN (TN>T0), all the normal related nodes have recorded the

cluster headâ€™s ID and then transmit their own monitoring data to the cluster head with

CSMA/CD protocol. If some related nodes are far away from the cluster head, they

should elect other related nodes as their relay nodes.

5) The cluster head gathers data from all other related nodes and performs data

aggregation. Moreover, the relay nodes inside the cluster also could perform data

aggregation if necessary.

As stated above, our algorithm ensures that the related node with more residual energy, nearer to the sink node and more connective with other nodes is selected as the cluster head. In addition, delay transmitting of CH-Select-M avoids the related nodes with low Pch broadcasting their own CH-Select-M, and thus reduces the number of overhead messages and energy waste.

3.3 Data Transmission

In LEACH and AEEC, the CHs transmit data packets to the sink node directly, that will be a good idea if the CH is near the the sink node. Otherwise, the CH will consume lots of energy for long-distance communication and go to dead in a short time, which will break down the connectivity of the network. In order to decrease the energy consumption of the CHs far from the sink node and balance the energy dissipation of the whole network, a new routing algorithm is presented. Firstly, we need to look back to the initialization of the network together with the process that each sensor node keeps record of its neighborsâ€™ information such as ID, residual energy, distance to the sink node and distance between each other. The following algorithm is based on those information.

1) Transmit the data packets to the sink node directly while the sink node is within the

neighbor records of the current node.

2) Otherwise, the current node chooses one of the next two principles to select the relay

node according to its own residual energy:

Principle 1: when Eres(c)<E0 and Eres(c)<Eaver, select the relay node by (8).

50 P m (i) =

E res (i) âˆ’ E aver

,

d (i) Ã— d (i)

âˆ€ i âˆˆ M

(8)

s c

max P m (i) â†’ next relay node

Principle 2: else, select the next relay node by (9).

http://mc.manuscriptcentral.com/tiisjournal

KSII Transactions on Internet and Information Systems Page 8 of 13

P m (i) = E res (i) / ds (i), âˆ€ i âˆˆ M

max P m (i) â†’ next relay node (9)

where Eres(c) is the residual energy of the current node. E0 is the initial energy of the node.

M is a set of current nodeâ€™s neighbors who are nearer to the sink node than the current

node. Eaver is the average energy of the current node and the nodes within M. ds(i) and dc(i)

are the distance between node i and the sink node and the distance between node i and the current node respectively.

Next, we present two explanations of the above principles.

Explanation 1: Multi-hop transmission consumes more energy than single-hop

transmission while it is a short trip, as shown in Fig.2. Thus, the current node should

transmit the data packets as far as it can while it has enough residual energy. As mentioned

above, all the nodes have set d0 as the default transmitting radius, which can achieve the

highest energy efficiency for unit transmission distance. Hence it will improve the energy efficiency of the whole network.

x 10-6

1.5

L=1

L=2

L=3

L=4

0.5

0 50 100 150 200

Transmitting distance(m)

Fig.2 The energy consumption of increasing transmitting distance with different hops

Explanation 2. Long distance transmitting should be avoided while the current node

has a little residual energy, so that the lifetime of the current node can be increased to a

certain extent. On the other hand, the relay routing should be almost straight between the

current node and the sink node, so as to reduce energy consumption for transmitting.

In addition, every sensor node should broadcast residual energy message when it

changes, and then the neighbors update the records accordingly. When a node happens to

have no live neighbors who are nearer to the sink node than itself under the current

transmitting radius, it needs to increase the power level to find an appropriate one just as

http://mc.manuscriptcentral.com/tiisjournal

Page 9 of 13

KSII Transactions on Internet and Information Systems

the way of network initialization.

4. Analysis and Simulation Results

In order to evaluate and analyze the performance of the proposed protocol, a simulation environment for wireless sensor network is established with MATLAB. And then we make some essential comparisons with AEEC and ARPEES. The simulation parameters are shown in Table 1.

Table 1. simulation parameters

Parameter Value

Monitoring area(MÃ—M) 500mÃ—500m

Number of sensors(N) 200

Initial energy of sensors(E0) 0.2J

Electronic energy(Eelec) 50nJ/bit

Energy for transmission in free space model(Îµfs) 10 pJ/bit/m2

Energy for transmission in multi-path model(Îµmp) 0.0013pJ/bit/m4

Energy of data aggregation(Edata) 5nJ/bit

Length of data packet(LdP) 200Byte

Length of control packet(LcP) 50Byte

Event related area(Revent) 50m

Fig.3 shows the total amount of residual energy of each round. EDCR keeps more residual energy than ARPEES and AEEC over each round and has a much longer lifetime compared to the other two methods. For AEEC, the total energy decreases sharply and goes to zero after 2500 rounds approximately. EDCR and ARPEES keep their own energy reducing slowly and reach a longer lifecycle about 11000 and 13000 rounds respectively. There are two main reasons for this difference. Firstly, EDCR and ARPEES apply event driven clustering mechanism while AEEC applies static clustering mechanism. And secondly, the cluster heads transmit data packets to the sink node directly in AEEC. However, EDCR and ARPEES apply single-hop or multi-hop transmitting according to the distance between the cluster head and the sink node, which avoids excessive energy consumption over long distance transmitting. In addition, EDCR shows a better performance over ARPEES from Fig.3, that is because EDCR takes the residual energy, the distance to the sink node and the connectivity with the neighbors into consideration when selecting the cluster head, while ARPEES chooses the node with maximum residual energy and nearest to the event as the cluster head. Besides, when choosing the relay node, ARPEES merely considers the residual energy and location of the candidate, and EDCR also takes the residual energy of the current node into account.

http://mc.manuscriptcentral.com/tiisjournal

KSII Transactions on Internet and Information Systems Page 10 of 13

EDCR

35 ARPEES

30 AEEC

0 5000 10000 15000

Number of event rounds

Fig.3 The total amount of residual energy per round with three methods

As mentioned, the lifetime of the first dead node has a significant impact on the whole

network. According to Fig.4, EDCR performs 10755-round activities before the first node

dies, while ARPEES and AEEC are 1141 and 1582 rounds respectively. Clearly, in EDCR

and AEEC, all the nodes get dead within a short duration while in ARPEES it continues for

a long time, and the curve is much rougher than the other two methods, which indicates

that ARPEES has a worse performance on energy consumption balance compared to

EDCR and AEEC.

200

EDCR

ARPEES

AEEC

150

100

0 0.5 1 1.5 2

Number of event rounds x 104

Fig.4 The number of living nodes per round with three methods.

Fig.5 and Fig.6 show the residual energy of each node after 2000 and 5000 rounds

http://mc.manuscriptcentral.com/tiisjournal

Page 11 of 13

KSII Transactions on Internet and Information Systems

respectively. Both illustrate that EDCR performs better than ARPEES and AEEC on energy economization and energy consumption balance.

0.3

EDCR

ARPEES

0.25

AEEC

0.2

0.15

0.1

0.05

0 50 100 150 200

Sensor node ID

Fig.5 The residual energy of each node after 2000 rounds with three methods.

0.25

EDCR

ARPEES

0.2

AEEC

0.15

0.1

0.05

0 50 100 150 200

Sensor node ID

Fig.6 The residual energy of each node after 5000 rounds with three methods

5. Conclusion

In this paper, an event driven clustering routing protocol (EDCR) for emergent event

monitoring applications of WSN was proposed. First of all, according to the radio model,

we found an optimal distance between the receiver and the transmitter which can obtain

the highest efficiency of energy consumption for unit transmission distance. We also

http://mc.manuscriptcentral.com/tiisjournal

Our Service Portfolio

Want To Place An Order Quickly?

Then shoot us a message on Whatsapp, WeChat or Gmail. We are available 24/7 to assist you.

Do not panic, you are at the right place

Visit Our essay writting help page to get all the details and guidence on availing our assiatance service.

Get 20% Discount, Now
£19 £14/ Per Page
14 days delivery time

Our writting assistance service is undoubtedly one of the most affordable writting assistance services and we have highly qualified professionls to help you with your work. So what are you waiting for, click below to order now.

Get An Instant Quote

ORDER TODAY!

Our experts are ready to assist you, call us to get a free quote or order now to get succeed in your academics writing.

Get a Free Quote Order Now

Transactions On Internet And Information Systems

= ï£²

,

Our Service Portfolio

Want To Place An Order Quickly?

Do not panic, you are at the right place

Get 20% Discount, Now £19 £14/ Per Page14 days delivery time

Get An Instant Quote

Get 20% Discount, Now
£19 £14/ Per Page
14 days delivery time