You are on page 1of 6

2015 International Conference on Advanced Computing and Communications

Uncertain Rule Based Fuzzy Logic QoS Trust Model in MANETs


Nageswararao Sirisala C.Shoba Bindu
Assistant Professor, CSE Dept Associate Professor , CSE Dept
Vardhaman College of Engg, JNTUACE
Hyderabad, India Anantapur, India
e-mail: nagsirisala@gmail.com e-mail: shobabindhu@gmail.com

2. RELATED WORK
Abstract- In Mobile ad-hoc networks (MANET), QoS
protocols includes the intermediate nodes which have 2.1 Ad Hoc QoS On-demand Routing (AQOR).
sufficient resources along the route, from source to AQOR[14] is the extension of AODV routing
destination. Nodes may turn non cooperative when
protocol with QoS awareness. Route discovery is
they are running with shortage of resources, which
weakens the effectiveness of the protocol. The
initiated when a route is required. The source node
proposed model (Uncertain Rule-based Fuzzy Logic sends a route request (RREQ) packets to all one
QoS Trust Model in MANETs -FQTM) selects the hop distance neighbour nodes, in which the
nodes which are capable and cooperative. Node bandwidth and delay requirements are included.
capability is evaluated through the QoS parameters( The intermediate nodes check their available
like energy, bandwidth, link stability) and resources. If the node is having enough bandwidth
cooperativeness is measured in terms of reliability. to handle the request, an entry will be created in the
Fuzzy logic is applied to compute the node trust value routing table with an expiration time. If the reply
by considering its quality and reliability metrics.
packet(RREP) is not received from destination
FQTM selects the nodes with higher trust values to
construct route from source to destination. The node in the allowed time, the entry will be
performance of proposed method is presented removed. So the reply packets, those are arriving
theoretically (time and space complexities) and beyond the timelines are not considered to avoid
experimentally. the overhead. After successful exchange of route
request and reply packets with threshold values of
Keywords: Fuzzy Logic; Quality of Service; Trust ; bandwidth and end-to-end delay, a source node
MANETs can establish the path to destination with adequate
bandwidth and time delay.
1 INTRODUCTION
In MANETs, there is no centralized 2.2 Trust metrics and Routing Protocols
administration, so the nodes may turn into selfish 2.2.1Trust metrics
when they are running shortage of resources. A
Trust can be reflected by reliability, utility,
malicious node can disgrace the effectiveness of
availability, reputation, risk, confidence, quality of
the routing protocol by dropping the source packets
services and other concepts. Trust can be computed
to destination. So the routing protocol has to
in different ways using different metrics. Some of
establish the route from source to destination
the significant trust metrics are discussed below.
through the intermediate nodes those are capable
Trust as belief function: In [11], Trust is
and cooperative.
represented as 3-dimensional metric as shown in
Some articles discussed Quality[6,8] and
figure1. Node As trust on node B is defined as
reliability[15] protocols separately. The proposed
method QoS Trust model considers nodes wBA (bBA , d BA , uBA ) , where the factors bBA , d BA
capability in terms of QoS metrics (like link A
and u B represents node As belief, disbelief and
bandwidth, node residual power, mobility) and
nodes cooperativeness in terms of reliability. uncertainty about node B , here the condition is
When a routing protocol is handling multiple bBA  d BA  uBA 1
QoS parameters it is not trivial to define priorities Uncertainity
1
to them, which depends on the current status of
nodes available resources. In some papers[3] this
issue is addressed with static weights to QoS 0 0
parameters, but may not be a worthy solution for
the applications with dynamic nature.
In this work, the proposed method FQTM uses
1
the fuzzy expert system to accomplish the balance 1 0 Disbelief
among the QoS parameters in evaluation of node Belief
trust value. It uses the intermediate nodes with high
Figure 1: trust triplet space
trust values for establishment of path between
source and destination.

978-1-4673-9777-3/16 $31.00 2016 IEEE 55


DOI 10.1109/ADCOM.2015.17
Trust as Probobility function: In some 3. SYSTEM MODEL
papers[12 ], probability is used as trust metric. A In this section, the estimation of QoS
node trust value is represented as the ration of parameters like node residual power, bandwidth,
number of packets it forwarded successfully to the link expiry time and reliability are discussed. Then
number of packets it received. A node can estimate uncertain rule base fuzzy logic to evaluate the node
its neighbour trust value as trust value is presented. Finally the proposed
T w( Rp ) u Rp  w( Rq ) u Rq  w( D) u D
routing protocol is presented.
Rps  Rpf Rqs  Rqf
Rp Rq
Rqs  Rqf 3.1 Nodes Residual Power
Rps  Rpf
Res  Ref In forwarding a k-bit data in d distance
Ds  D f
Re D coverage area, a node spends  energy[ 10]
Res  Ref Ds  D f
Etotal (k ) Eamp u k u d 2  2 u ( Eele u k )
here Rp, Rq, Re and D are Route reply, Route Where is the activation energy of transmitter
request, Route error and Data delivery or receiver and a amplifier has to spend 
misbehaviour factors. amount of energy in receiving and forwarding of
Trust as Fuzzy Logic: Trust is represented as data.
fuzzy sets[13] like very Low, Low, Medium and
High as shown in figure 2. Every fuzzy set defines 3.2 Link Band width
the domain values of trust. Fuzzy membership Node band width is estimated as number of free
functions are used to convert the trust crisp input slots[9] it is having with its downstream node. For
values into degree of membership values. Usually a node , a time slot can be scheduled as a
triangular membership values are used to represent transmission slot to its downstream neighbour , if
trust fuzzy sets. it is scheduled neither for sending nor receiving slot
Very Low Low Medium High at or  . If is one hop neighbour of , then
the same slot cant be utilized as a receiving slot
at node (converse is applicable for defining
receiving slot).
TSi {st S : st RSi , st TSi , st nkNB RSk } i
T
0 .33 .66 1 3.3 Node Mobility
Figure 2: triangular membership functions Nodes mobility is considered to estimate the
2.2.2 Trust-based source routing in mobile ad hoc link expiry time(LET) of a node to its downstream
networks[4] node[ 7]. Let A,B are two nodes those are in
When a sender node (Vs ) wants to share data to communication over a distance of d and ,
are their current positions respectively. If
the destination node (Vd), first it verifies its routing
cache information for valid route information with the nodes A,B are moving at velocities of and
high trust value. If such a route is found, it will along the directions of T1 and T 2 , then the
transfer the data packets through that valid route. communication duration (Link Expiry Time )
Otherwise it sends a FLOW-REQ message to all of inbetween A and B is defined as .
its one hop neighbours. FLOW-REQ message is LET
r 2 (a 2  c 2 )  (ad  cb)
2
 (ab  cd )
similarly RREQ(Route Reqest) packet in DSR (a 2  c 2 )

protocol. It contains the information like source ID, Where a v1 cos T1  v2 cos T2 , b x1  x2 ,
destination ID, sequence number, the list of c v1 sin T1  v2 sin T2 and d y1  y2
intermediate node Ids(Vi ,Vj) and their Trust values
through which it has come across. Upon receiving a 3.4 Estimation of Node reliability (cooperativeness)
message, a neighbour node forwards it to all next A nodes reliability is evaluated based on the
hop neighbours if the message is with high number of packets it is receiving and number of
sequence number of that source node or having the packets it is forwarding correctly. Using Bayesian
same sequence number but came across through the inference theory[2 ], a node reliability r is
route with higher trust value. On receiving multiple computed as degree of belief and its value lies in
FLOW-REQ messages, the destination node the range [0 1]. If a node received y number of
evaluates the RouteTV values for all possible packets and among them it forwarded x number of
routes to source and selects the better route with packets, then its reliability is estimated like.
high trust value. Destination node sends FLOW- f (r , y ) P( x / r , y )
SETUP message through the chosen path to source. f (r , y / x) 1

Source node starts the data transmission through


0
f (r , y ) P( x / r , y )dr

the established route to destination. Route trust Here the probability function  followes
value is computed as the binomial distribution and can be
RTVsd min ^TVij / Vi ,V j P, (Vi ,V j ) E`

56
described using Beta distribution. The above Uncertain Rule
equation can be represented like Base
E  y  x 1 D  x 1
( )(1  r )
y
x r
f (r , y | x) 1 1

r (1  r ) dr P( x / r , y) f (r , y)dr
D 1 E 1

0 0 Precise Inputs
Precise Output
i.e. f (r, y | x) # Beta( E  y  x, D  x) Energy,Bandwidth,
Fuzzyfication
Knowledge
Defuzzyfication Node Trust
Link expiry time & Base
And finally node reliability expectation function Reliability
Value
can be represented as
D
E[r ]
D E
Triangular
3.5 Fuzzy logic system Membership
functions
Fuzzy logic is a form of knowledge
representation suitable for notions that cannot be Figure 4: Fuzzy Logic in FQTM
defined precisely, but which depend upon their In the fuzzyfication process the expert system
contexts. Fuzzy logic is a expert system[1,5], which converts these crisp values into fuzzy values using
works based on fuzzy membership functions and rule base where all the rules are framed based on
rule base. nodes resources status. Here the input values(QoS
Fuzzification: Fuzzification is the process of metrics) are defined as four fuzzy sets like very
converting precise input quantity into fuzzy Low, Low, Medium and High using triangular
quantity using fuzzy membership functions. member ship functions as shown in figure 2 . In
Defuzzification: Difuzzification is the process of defuzzyfication the estimated node fuzzy trust
converting a fuzzy quantity into precise crisp value is converted into crisp value.
quantity. In this paper Centre of gravity (COG) is
4.2 Uncertain Rule Base:
used as a defuzzification method.
Rule base is the combination of rules those are
4. UNCERTAIN RULE BASED FUZZY QoS framed based on current availability of nodes
TRUST MODEL resources and network traffic. Every rule is in the
Network propoerties form of IF-THEN. Since each fuzzy input is
defined as 4 fuzzy sets, Rule Base can have
Node Free time Forwaded
Node Mobility
slots
Data size
packets
maximum of 256(4 ) rules. Some of the
(Direction&speed)
rules are represented in table1
4.3 Fuzzy Expert System:
QoS Metrics
In figure 5 , the node trust value is computed
LInk Expiry
Band Width Residual Energy Node Reliability using below two rules those are taken from rule
Time
base.
Rule 1: If(Reliability is Medium ,Energy is
Medium, LET is Low and Bandwidth is Medium)
Fuzzy Expert System then (Trust is Medium)

Fuzzy Logic
Rule 2: If(Reliability is Low, Energy is Low, LET
Precise Node Trust value is Medium and Bandwidth is Low) then (Trust is
Low)
Figure 3: Fuzzy Inference QoS trust model
Let the input precise values to fuzzy expert system
The proposed method is the extension of
are like Reliability = 0.498, Energy =36 J, LET=41
AODV, no additional control packets are used to
sec and Bandwidth =1.6 Mbps
gather network status information. As shown in
To find the fuzzy output value i.e node
Figure 3, by using HELLO packets a node can
trust value, max-min composition rule is applied,
collect all its one hop neighbour nodes information
which represents shaded area from both the rules as
like available free slots, velocity, moving directions
shown in figure 6.
and amount of source data. Based on this metrics a
node QoS metrics are estimated. Fuzzy expert 1
system uses QoS metrics as input values to
compute a node trust values. 0.5
4.1 Fuzzy inference system to evaluate node trust
values
In the proposed model, the fuzzy system considers 0 0.33 0.66 1
all the pre estimated QoS metrics as precise input Figure 6: Defuzzificaion of node trust value
values as shown in figure 4.

57
Medium Medium Low Medium Medium
1

Rule 1

20 60 J 0 60 1 0.33 1
1 3 0.33
sec
Mbps

Low Low Medium Low Low


1

Rule 2

2 90 0 0.66 0 0.66
0 40J 0 30
Mbps Sec
Energy
Bandwidth LET Reliability Trust

Figure 5: Example of node trust evaluation


TABLE 1: FQTM rule base

Ability Metrics Coopera


tiveness
Node Link Mobility Reliability Trust
Residual Bandwidth (LET)
Energy
High High High High High
High High Medium High High
Medium High Low High Medium
High Medium Medium High Medium
Medium Medium Low Very Low Low
Low Medium Medium Low Low
Very Low Low Medium Medium Very Low
Very Low Very Low Very Low Very Low Very Low
4.4Defuzzification: and computes trust value for every rule

In the defuzzification method, Center of i in rule base like
Gravity(COG) is used for converting fuzzy trust PE (Trust ) min{PC (b), P A (e), PD (et ), PB (r)} ,
i i i i i
value into precise trust value

  PE (Trust ) max{PE (Trust )} i


  In defuzzification, precise trust is
using above equation, the output trust value is calculate as  .
evaluated as 0.61 4) While forwarding RREQ packet to next hop
4.5 Algorithm of unicast routing in FQTM neighbour node adds its Id along with its
trust value  .
N is the number of nodes, SN is the set of
nodes. is an intermediate node. , are 5) If n j nd (destination node). nd collects all
destination and source nodes. , RREQ packets through different routes, for
, ,  , are the degree every route it evaluates Route trust(RT) as
of member ships of LET , energy, bandwidth, product of intermideate node trust values
reliability and trust metrics. f1, f2, f3, f4 represents  .
number of fuzzy sets of QoS metrics.
RT TV
Ni P , N j z N d
Nj

1) sends request packet RREQ to all one hop 6) nd sends the reply RREP through the route
neighbors nj with threshold QoS values. with higher trust value to the source. If two
2) If n j z nd (non destination node) , it routes are with same trust value then
estimates the available resources like r,e,b and minimum hop count route is preferred.
et for reliability , energy, bandwith and LET 7) On receiving Replay packet , source node ns
respectively . If these values are less than the establishes path to destination.
threshold value, then it ignore the RREQ 4.6 Time and Space Complexity of FQTM
packet.
3) If nj is having enough resources, through The proposed method doesnt use any additional
fuzzification it finds , , , control packets and using HELLO packets to gather

network information. It is spending O(m u n)

58
order of time to get neighbour node status, where n, In figure 8, throughput(number of packets
m are number of nodes and neighbour nodes. It reached at destination in unit time) is decreasing
requires O( p u m u n) amount of time in with the increasing node velocities. The proposed
method establishes the path with higher bandwidth
computation of neighbour node reliability, where p links, so it can attain the good delivery ratio.
is the number of packets. It needs O(n 2 ) order of
time in QoS metric computations. The rule base AODV AQOR FQTM
space complexity is O( f q ) , where f is number of
19
fuzzy sets, q is number of QoS parameters. 18
17

Throughput
16
5 SIMULATION RESULTS 15
14
The simulation is conducted in ns2 (2.34) tool. 13
Results are measured for the varying of network 12
11
size with 10 to 60 nodes and node velocity from 0 10
to 48 m/sec. The simulation area is 1200 u 1200, 9
8
node access region is 200m, transmission rate is 3
Mbps, initial energy of node is 60J and simulation 10 20 30 40 50 60
duration is 500sec. Simulation is run for the No of Nodes
AODV,AQOR and the proposed method FQTM.
Figure 9: number of nodes Verses control overhead.
In figure9, it is observed , throughput is
AODV AQOR FQTM
decreasing when the network size is increasing. If
0.75
Avg Packet Delay (Sec)

the network size is increased then node bandwidth


0.7 decreases. The proposed method let involves the
0.65
0.6 intermediate nodes with high bandwidth , so it can
0.55 out performs the remaining protocols.
0.5
0.45 6. CONCLUSION
0.4
0.35 The proposed method FQTM includes the
0.3 intermediate nodes which are capable and
0 6 12 18 24 30 36 42 48 cooperative. Uncertain rule base fuzzy logic is
Node Velocity m/sec applied to define the dynamic priorities of QoS
metrics, in computation of node trust value. FQTM
Figure 7: Node velocity Verses Avg packet delay could attain the significant improvement in packet
In figure 7, average packet delay(avg time delay end to end delay and throughput, since it is taking
taken by a packet to reach destination) is increased node mobility and energy into consideration. The
at higher node velocities . If node is moving with Performance of the proposed method is described
increased velocity, then the paths are getting
theoretically using asymptotic notations.
disconnected frequently which causes packet
buffering at intermediate nodes. The proposed REFERENCES
method considers node positions and their energy
[1] Fei Hao; Geyong Min; Man Lin; Changqing Luo; Yang,
levels, so it can reduce the path breaks and L.T. 2014, MobiFuzzyTrust: An Efficient Fuzzy Trust
improves the avg packet delay. Inference Mechanism in Mobile Social Networks, IEEE
Transactions on Parallel and Distributed Systems, vol.25,
no.11, pp.2944-2955.
AODV "AQOR" FQTM [2] Zhexiong Wei; Tang, H.; Yu, F.R.; Maoyu Wang; Mason,
P., 2014 Security Enhancements for Mobile Ad Hoc
21 Networks With Trust Management Using Uncertain
Reasoning, IEEE Transactions on Vehicular Technology,
Throughput

vol.63, no.9, pp.4647-4658.


16
[3] Sirisala, NageswaraRao.; C.Shoba Bindu. 2014 Weightage
based trusted QoS protocol in Mobile Adhoc Networks,
11 IEEE Global Conference on Wireless Computing and
Networking (GCWCN), pp.283-287.
[4] Lei Ju, Edwin H. M. Sha, Hui Xia, Zhiping Jia, Xin Li,
6 2013 Trust prediction and trust-based source routing in
0 6 12 18 24 30 36 42 48 mobile ad hoc networks Ad Hoc Networks Elsevier,Vol.
11, No. 7, pp. 2096-2114.
Node Velocity (m/sec) [5] Ben-Jye Chang and Szu-Liang Kuo, 2009 Markov Chain
Trust Model for Trust-Value Analysis and Key
Figure 8 : Node velocity Verses Throughput Management in Distributed Multicast MANETs, IEEE

59
Transactions on Vehicular Technology, vol.58, no.4,
pp.1846-1863.
[6] Mina Masoudifar, 2009 A review and performance
comparison of QoS multicast routing protocols for
MANETs Ad Hoc Networks 7(6), pp:11501155
[7] Nen-Chung Wang, Yung-Fa Huang Yu-Li Su 2007 A
Power-Aware Multicast Routing Protocol for Mobile
AdHoc Networks With Mobility Predictio, Wireless
Personal Communications, Vol 43:pp.14791497.
[8] Huayi Wu, Xiaohua Jia, 2007 QoS multicast routing by
using multiple paths/trees in wireless ad hoc networks, Ad
Hoc Networks Volume 5 Issue 5, pages 600-612.
[9] Kuei-Ping Shih. 2002 A TDMA-based bandwidth
reservation protocol for QoS routing in a wireless mobile ad
hoc Network, IEEE International Conference on
Communications Conference Proceedings ICC (Cat No
02CH37333) ICC-02.
[10] Heinzelman, W. R., Chandrakasan, A., & Baladrishnan, H.
2000. Energy-efficient routing protocols for microsensor
networks. Proceedings of the 33rd Hawaii international
conference on system sciences, Vol. 8, pp.110.
[11] G. Lenzini, M. S. Bargh and B. Hulsebosch, Trust-
enhanced security in location-based adaptive
authentication, Electronic Notes in TheoreticalComputer
Science, no. 197, pp. 105119, 2008.
[12] C. Zouridaki, B. L. Mark, M. Hejmo and R. K. Thomas,
Robustcooperative trust establishment for MANETs, in
The fourth ACMV workshop on security of ad hoc and
sensor networks, pp. 2334, 2006.
[13] J. Luo, X. Liu, and M. Fan, A trust model based on fuzzy
recommendation for mobile ad-hoc networks, Comput.
Netw., vol. 53, no. 14, pp. 23962407, 2009.
[14] Q. Xue and A. Ganz, Ad Hoc QoS On-demand Routing
(AQOR) in Mobile Ad hoc Networks, J. Parallel and
Distrib. Comp., vol. 62, no. 2, Feb. 2003, pp. 15465.
[15] Baolin, S. and Layuan, L. 2005 On the reliability of
MAODV in ad hoc networks, IEEE international
symposium on microwave, antenna, propagation and EMC
technologies for wireless communications, Vol. 2, pp.
15141517.

60

You might also like