International Journal of Computer Trends and Technology- volume3Issue2- 2012

Behaviour based Trust Management using geometric mean approach for Wireless Sensor
Networks
Ch.Satya Keerthi.N.V.L*1, A.Manogna#2, Ch.Yasaswini#2, A.Aparna#4 , S.Ravi Teja#5
* Lecturer, Department of Information science and Technology
#

Department of Information science and Technology,
KL University, Guntur – 522 502, Andhra Pradesh, India

Abstract — The resource constraints of Wireless Sensor
Network (WSN) make it easy to attack and hard to protect.
Although carefully designed cryptography and authentication help to make WSN securer, they are not good at dealing with compromised node and ageing node, whose misbehavior may impair the function of WSN. Hence, they are not sufficient for secure routing of message from source to destination in WSNs.
Alternatively Trust management schemes provide a powerful tool for the detection of unexpected node behaviours. A solution is obtained by first figuring out the malicious nodes in the network and then separating them from the benevolent nodes present on the basis of trust levels assigned. In this paper, we propose a new Trust Management System by considering the behaviours of sensor nodes, direct and indirect trusts based on
Geometric Mean (GM) of the QoS characteristics (trust metrics) among the nodes, which allows for trusted nodes only to participate in routing messages.
Keywords—Wireless Sensor Network (WSN), Geometric
Mean (GM), Trust Metrics, Quality of Services (QoS).

I. INTRODUCTION
Wireless Sensor Networks (WSNs) are ad-hoc networks, consisting of spatially distributed devices (motes) using sensor nodes to cooperatively monitor physical or environmental conditions at different locations. A sensor node is a tiny and simple device with limited computation and resources. Sensor nodes are randomly and densely deployed in sensed environment. WSN is designed to detect events or phenomena, collect and send back sensed data to the user. The power of wireless sensor networks lies in the ability to deploy large numbers of tiny nodes that assemble and configure themselves. Usage scenarios for these devices range from realtime tracking, to monitoring of environmental conditions, to ubiquitous computing environments, to in situ monitoring of the health of structures or equipment.
The characteristics of wireless infrastructure and characteristics of WSNs cause the potential risks of attacks on the network. Depending upon the application, WSN deployment environment may be hazardous, unattended and/or hostile and sometimes dangerous. The cryptographic,

ISSN: 2231-2803

authentication and other security mechanisms in WSNs cannot detect the node physical capture, the malicious or selfish nodes. Hence, they are not sufficient for secure routing of message from source to destination in WSNs ([1]-[3]). Also, the traditional cryptographic, authentication and other security mechanisms in WSNs consumes significant node resources and requires sophisticated software, hardware, large memory, high processing speed and communication bandwidth.
Another intricacy of the WSNs is that they operate in infrastructure less manner which further complicates the applicability of legacy security solutions [2].
Trust is defined as the subjective expectation a peer has about another’s future behavior based on the history of encounters. Trust [1] is “the degree of reliability” of other node in performing actions and can be formed by maintaining a record of the transactions with other nodes directly as well as indirectly. From this record a trust value will be established. Trust can be defined as “The subjective probability by which node A depends on node B to fulfil its promises in performing an action and at the same time being reliable in reporting its sensed data”. Trust is dependent on time; it can increase or decrease with time based on the available evidence through direct interactions with the node or recommendations from other trusted nodes. Trust-modelling is mathematical representation of node’s opinion of another node in a network. We need mathematical tools to represent trust and reputation, update these continuously [1].
Trust management system for wireless sensor networks
(WSNs) is a mechanism that can be used to support the decision-making processes of the network [4]. It aids the members of WSN to deal with uncertainty about the future actions of other participants (trustees). As WSNs are highly application oriented, these various applications bring various security needs. Survival of a WSN is dependent upon the cooperative and trusting nature of its nodes. Hence, the trust establishment between nodes is must.
Many researches on trust related in WSN are processed, but it is required to design and develop a light weight trust management system that takes the less resources of the node

http://www.internationaljournalssrg.org

Page 229

International Journal of Computer Trends and Technology- volume3Issue2- 2012 in calculation and management of trust between/among the nodes. The trust management of the WSN should be as simple as possible, i.e. without constraints on software, hardware, memory usage, computing, processing speed and communication bandwidth, and detect the different attacks easily, and mange and update trust relations accordingly.
A new approach of trust calculation considering both direct trust and indirect trust based on geometric mean (GM) of the
QoS characteristics such as packet forward, data rate, power consumption, reliability, etc. among the nodes is presented.
We have presented graphically, the trusted relations formed by all benevolent nodes of the WSN. Routing of data can take place through these benevolent or normal nodes present in the network thus reducing packet latency and dropping of packets.
Simulated results for different number of trust metrics for different trust threshold of this new trust management model are presented.
The paper is organized as follows. In Section II, we present the related work on WSN trust models. In section III, we describe the design of Geometric Mean based Trust
Management based on direct trust and indirect trust while in section IV we describe Behaviour based trust framework. We show the efficacy of the model through simulation results in section V. Lastly, we conclude the paper in section VI.

II. RELATED WORK –TRUST MODELS FOR SENSOR
NETWORKS.

neighbourhood, a reasonable mode is that every node maintains trust itself of neighbours while exchanging opinions with each other intermittently within the neighbourhood.
Mohammad Momani [1] proposed different methods for modelling and managing trust to enable WSN to be secure and reducing the computing and communication overheads. He proposed an algorithm for trust calculation and risk assessment based on trust factors and dynamic aspects of trust.
He modelled the direct trust computation with direct experiences, and indirect trust with recommendations given by the neighbours. The direct trust A of node N1 on node N2 is defined as the sum of trust values node N1 is having on node
N2 for different trust metrics such as packet forward, data rate, error rate, power consumption, reliability, competence, etc.
The indirect trust B of node N 1 on node N2 is defined as the average of recommendations given by the neighbours of node
N2 (nodes N3, N4, N5, N6, N7 as shown in Fig. 1. He modelled total trust using traditional weighting approach for direct trust and indirect trust as shown in following equations. A is direct trust (experience), B is indirect trust (recommendations), C is total trust. The direct trust A of N1 on N2 is given by the following equation.
=∑

Trust management can be distributed or central. In totally distributed mode, every node maintains its trust table of other nodes. Evidence of trust comes from direct observations of those nodes. In the central mode, every node reports its observations to a central node which acts as a trust authority in the WSN. While the former needs a long time to get the real estimation of a node’s trustworthiness, the latter invokes a quantity of communication and subsequent high energy cost.
Hybrid architecture may be much preferable than the former two. Because most interaction in WSN happens within

ISSN: 2231-2803

).

Sum of trust values of N 1 on N2 for m different trust metrics with different weights (Wi). The indirect trust B of node N1on node N2 is given by the following equation.
= ∗∑

Wireless Sensor Networks are categorized into three. They are centralized, hierarchical and distributed. Tae Kyung Kim, and Hee Suk Seo in [6] suggested a trust model using fuzzy logic for centralized WSN. Reputation defined as a perception of a party creates through past actions about its intentions and norms. The different components in the suggested trust modelling are minimum trust, maximum trust and un-trust.
They assumed that base station has the reputation value of each node. In most of the applications WSNs are either distributed or hierarchical, and hence the proposed model may not be suite for practical applications. Even if the WSN is centralized, they have not mentioned the way how base station evaluates the trust of a node.

(

∗

(

)

The indirect trust B of node N1 on node N2 is the average of direct trusts of k neighbours on N2.

=

∗

+

= ( , )
∗

The weights WA and WB can be assigned using different approaches. Some nodes may be given more weight for direct trust; others may be given more weight in indirect trust. i.e
WA > WB or WA < WB. Weights to the direct trusts of some metrics may be given more importance, and others are less importance. Similarly, for indirect trusts nearby nodes may be given more importance and others less. He introduced beta distribution system [7], [8] for weighting direct and indirect trusts in the case of communication trust. The trust management and modelling methods proposed by [1], are not light weight in consuming resources of node in the WSN.
In trust management scheme proposed in [9] honest nodes are favoured by giving them the credit for each successful packet forwarding, while penalizing suspicious nodes that exaggerate their contribution to routing.

http://www.internationaljournalssrg.org

Page 230

International Journal of Computer Trends and Technology- volume3Issue2- 2012

N3

Indirect information about node N2 given by its neighbors to node N1.
Direct information
N4
N5

N2

N6

N7

N1
Figure. 1. Node’s trust relationship.

Trust establishment system proposed in [10] has two ways to establish trust in computer networks. First, when the subject
(first party) can be directly observing the agent (second party).
Second, when the subject receives recommendations from other entities about the agent, indirect trust can be established.
A novel flexible trust management system proposed in [2] defined the trust as the ratio of successful transactions to total transaction made by the node. The proposed model is a decentralized trust scheme, i.e. the trust management functionality is distributed over the network nodes. In this model, each node is responsible for computing its own trust value per relation in the network, collecting direct and indirect information. The both direct and indirect trust values are used to evaluate each node’s trustworthiness. The proposed model has inherent reputation scheme of getting trustworthiness of any node, when direct evidences does not suffice, i.e. the number of direct evidences remain under threshold.

III. THE GEOMETRIC MEAN BASED TRUST
MANAGEMENT SYSTEM
In this section, we propose a new trust model suitable for many practical applications of the Wireless Sensor Networks
(WSNs). The concept here is each node in the network, will maintain a record for every its neighbour node. This record contains the information about different trust metrics, i.e. QoS characteristics for all its neighbours regarding the number of events occurred in the network. This trust metrics data will be helpful for calculating the direct trust of its every neighbour node. Also, as and when required, trust metric data of one node, can be transferred to other nodes, and act as indirect information for calculating indirect trust of nodes.
The proposed trust model is a decentralized trust scheme,
i.e. the trust management functionality is distributed over the network nodes. Each node is responsible for computing its own trust value per relation in the network, collecting events

ISSN: 2231-2803

from direct relations, and collecting trust values from other nodes in the network (in other words, indirect information).
This means that both direct and indirect trust values are used to evaluate each node’s trustworthiness. The indirect (secondhand) information may be particularly useful when no or limited direct interaction has been experienced.
One of the most important aspects of trust management schemes is the process of data collection for trust calculation.
The direct trust value of a neighbouring node can be determined by the different trust metrics of that particular node in different events occurred in the network. The trust metrics [11], i.e. the QoS characteristics that can be taken into account are shown in the Table 1.
TABLE 1. Trust Metrics
Data packets forwarded
Data packet/message precision
Control packet/ message forwarded
Control packet/ message precision
Availability based on beacon/hello messages
Routing protocol execution (routing actions)
Consistency of reported (sensed) values/data
Sensing communication
Reputation
Packet address modified

m1 m2 m3 m4 m5 m6 m7 m8 m9 m10 The listed trust metrics data for different events are essential and can provide a useful feedback to the system, towards the proper decision making by the trust management system. Here, depending on the application, we can insist the minimum level (threshold) to all the trust metrics, or we can have different thresholds to different groups of trust metrics.
Once one/more trust metric threshold/s are fixed, our trust management system see that no node is trusted unless the node is having minimum threshold level in a given trust metric strictly. This is the main advantage of our proposed trust management model comparing with other models.
The direct trust of any node, say on the node is a function of the all trust metrics values as shown in Fig. 2.
Here,
, ,…, are the trust metric values of on the node . These trust metrics tells node
, how node behaved in different QoS characteristics.

N2
DT = f (m1, m2, m3 , …, mn)

N1
Figure 2. Direct Trust function

In our trust management system, the direct trust is
Geometric Mean of all different trust metrics for different

http://www.internationaljournalssrg.org

Page 231

International Journal of Computer Trends and Technology- volume3Issue2- 2012 events occurred in the network on that particular node. These trust metrics are different from the trust metrics of other surrounding nodes. Like this, every node will be having a separate record of data of every surrounding node in different trust metrics for different events occurred in the network.
From these records, Direct Trust (DT) is calculated based on
Geometric Mean of the QoS characteristics as given in the below equations.
=
(

=

,

J, calculated for indirectly given information by L neighbours of J. This is shown in Fig. 4 clearly. The IT of node S on node
X is geometric mean of DTs of neighbours on node X. The total trust of any node with respect to any other node is again a function of Direct Trust (DT) and Indirect Trust (IT). Our proposed model also uses the traditional weighting approach for combining direct trust (DT) and indirect trust (IT) and form the total trust (TT) per relation in the network. As shown in following equation, DT is direct trust (experience), IT is indirect trust (recommendations), TT is total trust.

)

,…,

=

( )=

, ,

Here,
, ,…, are the trust metrics of node. The
( ) in the above equation is the Direct Trust value of node
I on node J, calculated for K different type of trust metrics.
Every node maintains the database of all its neighbours, and the contents of database are shown in Table 2.
The Indirect Trust on node

with respect to

can

TABLE 2. Node’s Database of its neighbour nodes
Nbr

k1 a1 k2

k3

...

kn

DT

IT

TT

TE

N1

a2

a3

...

an

DT1

IT1

TT1

TE1

N2

b1

b2

b3

...

bn

DT2

IT2

TT2

TE2

N3

c1

c2

c3

...

cn

DT3

IT3

TT3

TE3

N4

d1

d2

d3

...

dn

DT4

IT4

TT4

TE4

N5

e1

e2

e3

...

en

DT5

IT5

TT5

TE5

N6

f1

f2

f3

...

fn

DT6

IT6

TT6

TE6

N7

g1

g2

g3

...

gn

DT7

IT7

TT7

TE7

h1 h2 h3
...
hn
DT8
IT8
TT8 TE8
N8
Ni neighbours i=1 to 8, k1 ,k2 , ...,kn are type of trust metrics. DT direct trust. IT indirect trust, TT total trust and TE trust evaluation between two nodes. ai , bi , .., hi for i=1 to n are trust metric values for ki category for 8 neighbours be calculated from the direct trusts (DTs on with respect to its neighbours) sent by the neighbouring nodes of
. The neighbours of any node N are shown in Fig. 3. In our proposed model we have defined neighbour nodes to any node
N are N1, N2,N3, N4, N5,N6, N7, N8 in clockwise as shown in
Fig. 3. For example the neighbours of node N 2 are N3, N4, N5,
N6, N7 as per the Fig. 1. The Indirect Trust (IT) of node N1 on node N2 is defined as the Geometric Mean of the DTs of neighbour nodes (N3, N4, N5, N6, N7 as per Fig. 1.) on N2.
=
given by neighbour nodes.

= [∏(

,

( ) = [∏ (

,…,

= ( ,
+ ∗

∗

The weights is weightage given to DT and to the IT where +
= 1. Weights can be assigned using different approaches. Sometimes DT may be given more weight, and IT may be given less weight i.e.
>
.
( )=

( )∗

+

( )∗

The average Trust ( , ) between any two one hop nodes is given by the average of Total Trusts found each other as shown in Fig 5.
1
∗( ( )+
( ))
,
, =
2
This method provides many advantages as compared to the existing ones. It also allows us to find out the levels of trustworthiness of all neighbour nodes, even when the TETH value cannot be decided or evaluated, and hence the separation of benevolent and malicious nodes is possible. This method allows us to give more weight to certain trust metrics, depending on the requirement in the application. Calculation of direct trust of a single node with respect to another node based on number of parameters is accomplished by taking the average of the individual single parametric trusts in Momani’s model [1]. In Momani’s model, if the trust metric value for successful transmission of packets is 0 and the rest trust metrics have a high values, the overall trust value may be the above the trust threshold and the node will be trustworthy.
N8
N7

N1
N

N6

N2

N5

N3

information

)]

N4

( ))]

Here,
,
, …, are the DTs given by the neighbour nodes. The ( ) is the Indirect Trust value of node I on node

ISSN: 2231-2803

)

Figure 3. Node N and its neighbours N1, N2 , N3, N4,
N5, N6, N7, N8.

http://www.internationaljournalssrg.org

Page 232

International Journal of Computer Trends and Technology- volume3Issue2- 2012
N8

Trust Degree

N1

N7

DT1 , DT2 , DT3 , DT4 , DT5 , DT 6,
DT7 , DT8 .

DT1, DT2, DT3, DT4, DT5 , DT6 ,
DT7, DT8.

N6

N2

X
DT 1, DT2, DT 3, DT4, DT5, DT6,
DT 7, DT8.

Weight

Evaluation
DT1, DT2 , DT3 , DT4 , DT5 , DT6 , DT7 ,
DT8.

IT = [∏ (DT5 of N1 , DT6 of N2, DT7 of
N3, DT 8 of N4 , DT2 of N6, DT3 of N7 ,
1/7
DT4 of N8 )] .

S

Second hand information Combination

DT1 , DT2 , DT3 , DT 4, DT5, DT6,
DT7 , DT8 .

DT1 , DT2 , DT3 , DT4 , DT5, DT 6, DT 7,
DT8 .

Routing

I n p u t

Packet forwarding Other Tasks

N3

N4

O u t p u t DT1, DT2 , DT3 , DT4 ,DT 5, DT6, DT7, DT8.

Figure 4. Indirect Trust of node S on node X based geometric mean of direct trusts given by its neighbours.

That means it cannot detect the malicious/faulty nodes. But, our proposed trust model, this situation will be solved differently. The node will be treated as faulty or malicious because one of the trust metrics is failed to form trusty relation. In our trust model, trustworthy relation will be formed between two nodes only when the level of trust metric value for a given trust metrics is greater or equal to trust metric threshold.
IV. BEHAVIOUR BASED TRUST FRAMEWORK
The process of the behaviour-based trust is depicted in figure 5.After every task, the behaviour of the node is evaluated. The evaluation result is combined with old trust degree to form a new one. The new degree is considered in the next task allocation in term of weight. For example, the data from node with high degree should play a more important role in data fusion, and the trustworthy nodes should be chosen to transmit data with higher probabilities than untrustworthy ones. To calculate the trust degree here the geometric mean based Trust management system is used.
Nodes with their previous activities and behaviour patterns are distinguished as reliable nodes and unreliable nodes.
Reliable nodes are nodes with high confidence level and unreliable nodes are nodes with low confidence level, i.e., to say that nodes crossed the threshold confidence level are reliable, which have confidence level value less than that are unreliable. As this is a distributed system [5], each node has the confidence level values of its immediate neighbours. So, it may turn out be an unreliable node for one node might be reliable for another node. Every node maintains confidence level matrix of its immediate neighbours, which are later required for trust management.

ISSN: 2231-2803

Figure. 5. The process of behavior-based trust management

V. PERFORMANCE RESULTS
The performance of the proposed trust management system has been evaluated through computer simulations. A new software simulation package has been developed using the
MATLAB platform to model our approach. A Wireless Sensor
Network with 30 nodes is placed in the environment based on the locations in the coordinate system. The figure 6 shows the random network deployed. All the nodes in the network are self organising i.e, ad-hoc. They are randomly placed based on the coordinate locations. Once the nodes verify their locations they check for their neighbours based on their communication range. In our simulated network the range of communication for each node is 200. The neighbours of the nodes are represented using the connecting lines between them.
The trust values computed through the geometric mean approach are used for obtaining the best path in routing the data to the internet through the gateway. All the neighbours of the nodes and their trusts are stored in the form of a three dimensional matrix in MATLAB. For instance the node 1 in the simulated network has node 2 and 10 as its neighbours.
The corresponding data table at node 2 will store the trust of node 1 in the perspective of node 2 computed through the behaviour based trust framework. To obtain the effective trust of node 1 we need to find the average of the trusts at all its neighbours. From the graphical representations of trust values among the WSN nodes as shown in Fig. 7, we can see how the trusts of nodes vary with the tasks and their performance. The node
31 in the graph is the gateway which has the highest trust value so that all the nodes in its neighbourhood can communicate with it to transfer the data to the base station.

http://www.internationaljournalssrg.org

Page 233

International Journal of Computer Trends and Technology- volume3Issue2- 2012
The results shown are for randomly taken trust metrics only, not for practical live metric values.
VI. CONCLUSIONS
The trust management system presented provides the trust relations among the nodes depending on the QoS characteristics in the network. In this proposed paper, we have concentrated to form the trust evaluation between two nodes based on weightage of different trust metrics i.e.
QoS characteristics, weightage of direct trust, and for overall level of trust between nodes (level of trust evaluation). We combined the geometric mean approach with the behaviour based framework for improving the reliability of the simulated environment. We deployed a
WSN and obtained the trust values of the nodes in the dynamic environment. Though our proposed model is computationally simple, our future work includes finding communication overhead and modelling other malicious behaviour patterns to make the distributed system more reliable.Also, we plan to find the best trusted route among the many trusted routes from source to destination by applying our proposed model.
References:
[1]

Figure 6. Wireless Sensor network with ad-hoc placement of nodes

Figure 7. Trust values for nodes in the simulated network

ISSN: 2231-2803

M. Momani, “Bayesian Methods for Modeling and Management of
Trust in Wireless Sensor Networks,” Ph.D. Thesis, University of
Technology, Sydney, July, 2008.
[2] P. Trakadas, S. Maniatis, T. Zahariadis, H.C. Leigou, S. Voliotis, “A novel flexible trust management system for heterogeneous sensor networks”, an International Symposium on Autonomous
Decentralized Systems, 2009, proceedings ISADS 2009, page(s):369-374. [3] Mayank Saraogi; Security in Wireless Sensor Networks; Department of Computer Science, University of Tennessee, Knoxville.
[4]
Javier Lopez, Rodrigo Roman, Isaac Agudo, Carmen FernandezGago, “Trust management systems for wireless sensor networks:
Best practices”, Computer Communications, Elsevier, 33(2010) pp.
1086 – 1093.
[5] Arijit Ukil “Secure Trust Management in Distributed Computing
Systems”, 2011 Sixth IEEE International Symposium on Electronic
Design, Test and Application,116-121. doi:10.4236/wsn.2011.34015. [6] Tae Kyung, and Hee Suk Seo, “A Trust Model using Fuzzy Logic in
Wireless Sensor Network”, World Academy of Science, Engineering and Technology 42 2008.
[7] Mohammad Momani, Khalid Aboura, Subhash Challa,
“RBATMWSN: Recursive Bayesian Approach to Trust Management in Wireless Sensor Networks”, IEEE, ISSNIP 2007.
[8] Mohammad Momani, “Bayesian Fusion Algorithm for Inferring
Trust in Wireless Sensor Networks”, JOURNAL OF NETWORKS,
VOL 5, NO 7, JUL 2010. pp 815-822.
[9] K. Liu, N. Abu-Ghazaleh, and K.-D. Kang, “Location Verification and Trust Management for Resilient Geographic Routing,” J. Parallel and Distributed Computing, vol. 67, no. 2, pp. 215-228, 2007.
[10] Yan Sun, Zhu Han, and K. J. Ray Liu, “Defense of Trust
Management Vulnerabilities in Distributed Networks”, IEEE
Communications Magazine, Feature Topic on Security in Mobile Ad
Hoc and Sensor Networks, vol. 46, no. 2, pp.112-119, February
2008.
[11] Shaik Sahil Babu, Arnab Raha, Mrinal Kanti Naskar “Geometric
Mean based Trust Management System for WSNs (GMTMS)”, 2011
World Congress on Information and Communication Technologies.

http://www.internationaljournalssrg.org

Page 234