Performance Evaluation of Energy and Delay Aware Quality of Service (QoS) Routing Protocols in Mobile Adhoc Networks

Abstract

Mobile adhoc network (MANET) is a collection of mobile devices which form a communication network with no pre-existing wiring or infrastructure. Multiple routing protocols have been developed for MANETs. As MANETs gain popularity, their need to support real time applications is growing as well. Quality of service(QoS) provisioning is becoming a critical issue in designing mobile adhoc networks due to the necessity of providing multimedia applications.These applications have stringent QoS requirements such as throughput, end-to-end delay, and energy. Due to dynamic topology and bandwidth constraint supporting QoS is a challenging task. QoS aware routing is an important building block for QoS support. The primary goal of the QoS aware protocol is to determine the path from source to destination that satisfies the QoS requirements. This article proposes a new energy and delay aware protocols called, energy and delay aware Adhoc On demand Distance Vector Routing (EDAODV) and energy and delay aware Dynamic Source Routing(EDDSR) based on extension of AODV and DSR. Simulation results show that the proposed protocols have a better performance than AODV and DSR in terms of energy, packet delivery ratio and end-to-end delay.

Full text

437
Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
ABSTRACT
Mobile adhoc network (MANET) is a collection of
mobile devices which form a communication net-
work with no pre-existing wiring or infrastructure.
Multiple routing protocols have been developed
for MANETs. As MANETs gain popularity, their
need to support real time applications is growing
as well. Quality of service(QoS) provisioning is
becoming a critical issue in designing mobile
adhoc networks due to the necessity of providing
multimedia applications.These applications have
stringent QoS requirements such as throughput,
end-to-end delay, and energy. Due to dynamic
topology and bandwidth constraint supporting
QoS is a challenging task. QoS aware routing
is an important building block for QoS support.
The primary goal of the QoS aware protocol is
to determine the path from source to destination
thatsatisestheQoSrequirements.Thisarticle
proposes a new energy and delay aware proto-
cols called, energy and delay aware Adhoc On
demand Distance Vector Routing (EDAODV)
and energy and delay aware Dynamic Source
Routing(EDDSR) based on extension of AODV
and DSR. Simulation results show that the pro-
posed protocols have a better performance than
AODV and DSR in terms of energy, packet de-
livery ratio and end-to-end delay.
Chapter 2.8
Performance Evaluation of
Energy and Delay Aware Quality
of Service (QoS) Routing
Protocols in Mobile Adhoc
Networks
R.Asokan
Kongu Engineering College, India
A.M.Natarajan
Kongu Engineering College, India

438
Performance Evaluation of Energy and Delay Aware Quality of Service (QoS)
INTRODUCTION
Wireless mobile networks and devices are be-
coming increasingly popular as they provide
users access to information and communication
anytime, anywhere. Conventional wireless mobile
communication is usually supported by a wired
xedinfrastructure,suchasasynchronous transfer
mode (ATM) or the Internet. The mobile devices
use single-hop wireless radio communications
to access a base station that connects the wired
infrastructure. In contrast, the class of mobile ad
hocnetworks(MANETs)doesnotuseanyxed
infrastructure. The nodes of MANETs intercom-
municate through single-hop and multihop paths
in a peer-to-peer fashion. Intermediate nodes
between a pair of communicating nodes act as
routers. Thus, the nodes in MANETs operate as
both hosts and routers. The nodes are mobile and
so the creation of routing paths is affected by the
addition and deletion of nodes. This results in the
rapid change in topology of the network.
The proactive table-based routing schemes
require the storage of the routing information,
which is used to determine the next hop for the
packet transmission to reach the destination. The
protocol attempts to maintain the table informa-
tion consistent by transmitting periodical updates
throughout the network. These routing schemes
maybeatorhierarchicalinnature.Examples
of flat table-based routing schemes include
destination-sequenced distance vector (DSDV)
routing and wireless routing protocol (WRP). Flat
routing schemes require maintenance of the state
of the entire network at all nodes, which limits its
scalability. In the hierarchical approach, the state
of only a subset of the network is maintained at
all nodes, and routing is facilitated through an-
other level of state information, which is stored
in fewer nodes.
In the case of on-demand source-based routing
schemes, routes are created as and when neces-
sary based on a query-reply approach. When a
node needs to communicate with another node, it
initiates a route discovery process. Once a route
is found, it is maintained by a route maintenance
procedure until the route is no longer needed.
Examples of on-demand source-based routing
schemes include ad hoc on-demand distance vector
(AODV) routing protocol, dynamic source routing
(DSR), and the temporary ordered routing algo-
rithm (TORA) (Royer & Toh,1999). These algo-
rithmsfocusonndingtheshortestpathbetween
the source and destination nodes by considering
thenodestatusandnetworkcongurationwhen
a route is desired.
Quality ofservice (QoS)is usually dened
as a set of service requirements that needs to be
met by the network while transporting a packet
stream from a source to its destination.QoS rout-
ing protocols search for routes with sufcient
resources in order to satisfy the QoS require-
mentsofaow.TheQoSroutingprotocolshould
ndthepaththatconsumeminimumresources
(Prasant, Jian, & Chao, 2003). Depending on the
application involved, the QoS constraints could
be available bandwidth, cost, end-to-end delay,
delay variation (jitter), energy, probability of
packet loss, and so forth.
The rest of the article is organized as follows.
In the next section, the previous work related to
QoSawareroutingprotocolsisbrieyreviewed.
Subsequently, energy and delay aware protocols
called energy and delay aware AODV (EDAODV)
and energy and delay aware DSR (EDDSR) based
on extension of AODV and DSR are described.
Following that, the major simulation results are
shown. Finally, the result of the work done is
summarized.
RELATED WORK
The primary goal of the QoS-aware routing pro-
tocols is to determine a path from a source to the
destinationthatsatisestheneedsofthedesired
QoS. The QoS aware path is determined within
the constraints of bandwidth, minimal search,

439
Performance Evaluation of Energy and Delay Aware Quality of Service (QoS)
distance, and trace conditions. Because the path
selection is based on the desired QoS, the routing
protocol can be termed as QoS-aware. Only a few
QoS-aware routing protocols have been proposed
so far for MANETs, most of which are outlined
in this section.
A power-aware multiple access protocol (PA-
MAS) has been proposed (Singh & Raghavendra,
1998). Here, a node turns off its radio interface for
aspecicdurationoftime,whenitknowsthatit
will not be able to send and receive packets during
that time because of the possibility of multiple ac-
cess interference. The sleep time is of the order of
packet duration which could be very small. This
approach would be quite viable for low bandwidth
mobile networks. Several energy-aware metrics
thatwillresultinenergy-efcientrouteshavebeen
discussed (Singh, Woo, & Raghavendra, 1998a). It
is hard to use these metrics directly in a network
without any central control. The performance
evaluation compares energy aware cost model
with the shortest path routing.
Routing algorithm based on minimizing the
amount of power (or energy per bit) required has
been proposed to get a packet from source to des-
tination (Singh, Woo, & Raghavendra, 1998b).
Route selection depends on the packet size,
and hence, in the case of variable packet size
transmission, many routes should be selected.
Conditional Max-Min battery capacity routing
algorithm has been proposed (Toh, 2002).This
algorithm chooses the route with minimal total
transmission power if all nodes in the route have
remaining battery capacities higher than a thresh-
old; otherwise routes including nodes with the
lowest remaining battery capacities are avoided.
Theresultsshowedthattherstnodeinshortest
path routing metric died sooner than all the bat-
tery cost aware routing.
Sivakumar, Sinha, and Bhargavan (1999) have
proposed the Core-Extraction Distributed Ad Hoc
Routing (CEDAR) algorithm. CEDAR is designed
toselectrouteswithsufcientbandwidthresourc-
es. CEDAR dynamically manages a core network,
on which the state information of those stable high
bandwidth links is incrementally propagated. Each
core node is responsible for maintaining its local
topology as well as calculating routes on behalf of
nodes in its vicinity. CEDAR selects QoS routes
upon request. The performance of CEDAR largely
depends on how well core nodes can manage their
local resources.
Lin and Liu (1999) have proposed an available
bandwidth calculation algorithm for ad hoc net-
works with time division multiple access (TDMA)
for communications. This algorithm involves
end-to-end bandwidth calculation and bandwidth
allocation. Using this algorithm, the source node
can determine the resource availability for sup-
porting the required QoS to any destination in the
ad hoc networks. This approach is particularly
suitable in call admission control.
Liao, Tseng, Wang, and Sheu (2001) have pro-
posed a multipath QoS routing protocol which is
also an extension of DSR. Unlike other existing
protocolsforadhocnetworks,whichtrytond
a single path between source and destination, this
algorithm searches for multiple paths for the QoS
route, where the multiple paths refer to a network
with a source and a sink satisfying certain band-
width requirement. The multiple paths together
satisfy the required QoS. It is suitable for adhoc
networks with very limited bandwidth where a
single path satisfying the QoS requirements is
unlikely to exist.
ENERGY AND DELAY AWARE
PROTOCOLS
The QoS metrics can be classied as additive
metrics, concave metrics, and multiplicative met-
rics. Let m(u,v) be the performance metric for the
link (u,v) connecting node u to node v, and path
(u,u1,u2…uk,v) a sequence of links for the path from
u to v. A constraint is additive if m(u,v) = m(u,u1)
+ m(u1,u2) +...+ m(uk,v).The end-to-end delay is an

440
Performance Evaluation of Energy and Delay Aware Quality of Service (QoS)
additive constraint because it is the accumulation
of all delays of the links along the path.
A constraint is concave if m(u,v) = min{m(
u,u1),m(u1,u2),...,m(uk,v)}.The bandwidth bw(u,v)
requirement for a path between node u and v is
concave.TondaQoSfeasiblepathforacon-
cave metric, the available resource on each link
should be at least equal to the required value of
the metric.
A constraint is multiplicative if m(u,v) =
m(u,u1) x m(u1,u2) x ... x m(uk,v). The probability
of a packet prob(u,v), sent from a node u to reach a
node v, is multiplicative, because it is the product
of individual probabilities along the path. Band-
width and energy are concave metric, while cost,
delay, and jitter are additive metrics. Bandwidth
and energy are concave in the sense that end-
to-end bandwidth and energy are the minimum
of all the links along the path. The reliability or
availability of a link based on some criteria such
as link break probability is a multiplicative metric
(Baoxian & Hussein, 2005).
Delay
The delay is the total latency experienced by a
packet to traverse the network from the source to
the destination. At the network layer, the end-to-
end packet latency is the sum of processing delay,
packetization, transmission delay, queuing delay,
and propagation delay. The end-to-end delay of a
path is the summation of the node delay at each
node plus the link delay at each link on the path.
Node delay includes the protocol processing time
and the queuing delay at node i for link (i,j). Link
delay is the propagation delay on link (i,j).
Shen and Chen (2001) analyzed the relation-
ship between the medium access control (MAC)
delay and the neighbor number in mobile ad hoc
networks, and provided an estimation method of
the MAC delay. Sun and Hughes (2003) analyzed
queuingdelaybyusingtwodimensionnite-state
Markov model. They gave the queuing delay
distribution Pr (D >t) .The average queuing delay
isdenedtobethevalueD for which the delay
distribution is larger than 90%. Thus, the end-to-
end delay of a path can be estimated by adding all
the node delays and link delays in the path.
Energy
On-demand protocols typically pick the shortest
path route during the route discovery process and
then stick to this route until it breaks. Continuous
use of the route may drain the energy of the nodes.
This is particularly true if one or more nodes are
on other routes as well. Each message transmis-
sion and reception drains the battery power. If a
node runs out of battery energy and is unable to
forward any messages, it effectively falls out of
the network. In this case, the route breaks and
protocolndanalternaterouteviaanotherroute
discovery. Dying of nodes will affect the opera-
tional life time of the ad hoc network.
Mobile devices generally depend on nite
battery sources, so QoS provisioning must
consider residual battery power. The aim of the
protocol is routing around nodes high on battery
power as far as possible. This will prolong the
network lifetime. The percentage of the initial
energy is taken as the energy metric in the QoS
specication.Theminimumenergyisselectedby
the node which initially requests the route. The
application and duration of transmission are the
factors that determine the minimum energy. It is
assumed that the initial energy of the node is the
maximum energy provided by the battery when
it is fully charged.
Energy and Delay Extension in
AODV and DSR
Theminimumenergyandmaximumdelayelds
are added with the routing table for each destina-
tion. A source requiring minimum energy and
maximum delay transmits a route request (RREQ)
packet with QoS energy and delay extension as
shown in Figure 1. The energy extension indicates

441
Performance Evaluation of Energy and Delay Aware Quality of Service (QoS)
the minimum energy required to be available on
the entire path between the source and destination.
The extension of delay gives the maximum delay
allowed between the source and destination.
As shown in Figure 1, the QoS energy exten-
sion is 30% (0.3) of the node’s initial energy and
the maximum delay is 100 milliseconds (ms).
Both minimum energy and maximum delay
vericationsofRREQhavebeen doneineach
node. RREQ packets are discarded if one of the
constraintscannotbesatised.
Before forwarding the RREQ packet an in-
termediate node compares its available energy
totheenergyeldindicatedintheQoSexten-
sion. If the required energy is not available, the
packet is discarded and the process is stopped. If
theenergyconstraintissatised,thenthedelay
is estimated and if it exceeds the QoS delay the
packet is discarded; otherwise the node subtracts
its node traverse time (NTT) from the delay
bound provided in the extension. The delay value
in RREQ packet indicates the delay allowed for
a transmission between the source and destina-
tion. The RREQ is forwarded with updated QoS
delay extension.
The delay in RREP packet indicates the
estimate of the cumulative delay allowed for
a transmission between the intermediate node
which forwards the RREP and destination. In
response to the QoS request (RREQ), the desti-
nation sends an RREP packet with its measured
available energy and initial delay corresponding
S A D B
0.7, 20 ms 0.8, 30 ms
RREQ, S, D, 0.3, 100ms RREQ, S, D, 0.3, 80ms RREQ, S, D, 0.3, 50ms
0.8, 10 ms
Figure 1. QoS route request for energy and delay
to its NTT. Each intermediate node forwarding
theRREPcomparestheenergyeldoftheexten-
sion with its own available energy on the selected
route and keeps the minimum between these
two values to propagate the RREP, as shown in
Figure 2. This value is recorded in the routing
table for the destination. In the case of delay the
intermediate node adds its own NTT to the delay
eldandrecordsthisvalueintheroutingtablefor
the concerned destination before forwarding the
RREP. This entry update allows an intermediate
node to answer the next RREQ by comparing the
maximumdelayeldinthetable.
Theowchart,asshowninFigure3,describes
the sequence of operation. Energy and delay
metrics are used in AODV route discovery. Each
RREQpacketoodedinthenetworkbuildsup
the cost for the path traversed so far by the packet.
Each routing table entry also maintains energy
and delay for that route. In regular AODV the
nodeactsononlytheveryrstRREQreceived
per route discovery. Duplicates of the RREQ
received via alternate routes are ignored. How-
ever, use of these new cost metrics requires that
AODV acts on all such duplicates if they carry a
lower cost metric. If a RREQ arrives with a lower
cost metric, it is forwarded when the node is not
the destination and does not have a route to the
destination, and otherwise it is replied. Energy
and delay metrics are used as QoS extensions in
DSR route discovery. Every node receiving this
route request searches through its route cache

442
Performance Evaluation of Energy and Delay Aware Quality of Service (QoS)
for a route to the requested destination with the
specied energy and delay. Each node’s route
cache will have the energy and delay values. If
bothenergyanddelayconstraintissatised,then
the intermediate node forwards the RREQ to the
next node; otherwise, it is discarded.
PERFORMANCE EVALUATION
The performance of the proposed protocols is
evaluated using the ns-2 simulator (Eitan &
Tania, 2003). In the simulation, the ns-2 Wave-
LAN implementation for MAC 802.11 is used.
This MAC implementation uses 2 Mbps as
channel access rate. The simulation is done for
a network of 50 mobile nodes with a maximum
speed of 70km/hour. Each node is moving in an
area of 670x670 m2. The node radio transmission
range is about 250m. The pause time is varied to
simulate different degrees of mobility. A longer
pausetimemeansalowermobilityprole.The
QoS constraint is set to 250 ms and 350 ms for
delay and 30% of initial energy for energy. The
initial energy for each node is set to 20 J, which
represents a combined network-wide initial energy
of 1000 J. Table 1 lists the simulation parameters
and environments used.
Performance Metrics
The following metrics are used to evaluate the
performance of the protocol.
Figure 2. QoS route reply for energy and delay
S A
D
B
0.7, 20 ms 0.8, 30 ms
RREP, S, D, 0.7, 60ms RREP, S, D, 0.8, 40ms RREP, S, D, 0.8, 10ms
0.8, 10 ms
• Packet delivery ratio: Measured as the ratio
of the number of data packets delivered to the
destination and the number of data packets
sent by the source.
• End-to-end delay: It indicates the delay
experienced by packets from source to des-
tination. This delay includes processing and
queuing delays in each intermediate node.
• Remaining energy: It is measured as the
total amount (in Joules) of remaining battery
energy at the end of the simulation.
Packet Delivery Ratio
Figure 4 shows the effect of mobility on packet
delivery ratio (PDR) of four protocols with transfer
of 20 packets and 60 packets per second. Dur-
ing high mobility, all the protocols show small
degradation in PDR due to high link breakage.
EDAODV shows 4% to 8% improvement in PDR
in 20 packets and 9% to 16% improvement in the
60 packets transfer. EDDSR shows up to 10%
improvement in packet delivery ratio in both the
cases over DSR. EDAODV and EDDSR show
better performance under moderate and high
mobility conditions. The improvement is more
signicantinhigherpackettransfer.
Figure 5 shows effect of pause time on PDR
of the four protocols. The PDR tends to increase
as the pause time increases, because active path

443
Performance Evaluation of Energy and Delay Aware Quality of Service (QoS)
is less likely to break as the network becomes
static. EDAODV and EDDSR show slightly better
performance at low pause time.
End-to-End Delay
Figure 6 shows the effect of mobility on end-to-
end delay for two QoS requirements 250 ms and
350 ms. The end-to-end delay increases as the
Figure 3. Flow chart
No
Yes
No
Node energy
>=QoS
Energy
RREQ
No Is the next
node
destination
Forward RREQ
to the next node
Estimated
delay <=
QoS delay
Initiate
RREP
Terminate
the RREQ
Yes
Yes
node speed increases. Higher mobility causes
more broken links and frequent rerouting and
thus causes more packet loss and larger end-to-
end delay.
The end-to-end delay is about 75 ms less in
250 ms and 80 ms less in 350 ms QoS delay for
EDAODV. EDDSRsatises therequirementin
250 ms and 30 ms less delay in 350 ms. It is ob-
servedthattwoQoSrequirementsaresatised
in EDDSR and EDAODV.

444
Performance Evaluation of Energy and Delay Aware Quality of Service (QoS)
Figure 5. Effect of pause time on packet delivery
ratio
Remaining Energy
Figure 7 shows effect of pause time on remain-
ing energy under four protocols. The remaining
energy at the end of simulation is much higher
for EDAODV and EDDSR than AODV and DSR.
In EDAODV the improvement is about 8 times
for low pause time and up to 5 times for high
pause time.
In the case of EDDSR, the improvement is
about 60 times at low pause time and 6 times at
high pause time. However, these improvements
Number of nodes 50
Terrain range 670*670 square meter
Transmission range 250 meter
Mobility model Random way point model
Speed 0-70 km/hr
Pause time 0 to 900 seconds
Propagation model Free space
Channel data rate 2 Mbps
Medium access control (MAC) 802.11
Table 1. Simulation parameters
Figure 4. Effect of mobility on packet delivery ratio
(a) 20 packets per second (b) 60 packets per second
strongly depend on the initial energy and the
simulation time.
Figure 8 shows the time at which a certain
number of nodes die, when simulating four pro-
tocols. It can be observed that nodes in AODV
and DSR protocols die earlier than EDAODV and
EDDSR protocols. This is due to forwarding the
RREQ packet, the intermediate node compares
itsavailableenergytotheenergyeldindicated
in the QoS extension. If the required energy is
notavailablethe packetis discarded. The rst
node in AODV dies 1,500 seconds earlier than
EDAODV,andsimilarlythe rstnodein DSR
dies 1,000 seconds earlier than EDDSR. EDAODV
and EDDSR increase the network lifetime by 27%
and 22%, respectively.

445
Performance Evaluation of Energy and Delay Aware Quality of Service (QoS)
Figure 6. Effect of mobility on end-to-end delay
(a) 250 ms QoS delay (b) 350 ms QoS delay
Figure 7. Effect of pause time on remaining
energy
Figure 8. Time vs. number of dead nodes
CONCLUSION
Energy and delay aware protocols EDAODV
and EDDSR based on extension of AODV and
DSR have been proposed in this article. At route
discovery phase, nodes that do not have required
energy and delay are eliminated. Each node upon
receipt of the RREQ packet determines whether
to forward this request based on its energy level
and delay or not. At the destination, route reply
is generated only for the request with minimum
routing cost. Simulation results show that both
protocols satisfy the energy and delay QoS re-
quirements. In our simulation study it has been
found that both the EDAODV and EDDSR out-
perform the AODV and DSR protocols in terms
of energy, packet delivery ratio and end-to-end
delay. So these protocols can provide excellent
energy and delay assurance, and at the same time
achieve much higher packet delivery ratio than
the existing protocols. In the future this can be
extended to other routing protocols. These energy
and delay aware protocols work only in the routing
layer and exploit only route-specicinformation.
The use of MAC layer information, transport and

446
Performance Evaluation of Energy and Delay Aware Quality of Service (QoS)
application layer information can also be explored
in further studies.
REFERENCES
Baoxian, Z., & Hussein, T. M. (2005, October).
QoS routing for wireless ad hoc networks: Prob-
lems, algorithms, and protocols. IEEE Commu-
nications Magazine, 110-117.
Eitan, A., & Tania, J. (2003, December).NS
simulator for beginners, lecture notes. Retrieved
August 16, 2007, from//www.isi.edu /nsnam/ns/
ns-documentation.html
Liao, W.H., Tseng, Y.C., Wang, S.L., & Sheu,
J.P. (2001). A multi-path QoS routing protocol in
a wireless mobile ad hoc network. In Proceed-
ings of the IEEE International Conference on
Networking, (pp. 158-167).
Lin, C.R., & Liu, J.S. (1999, August). QoS rout-
ing in ad hoc wireless networks. IEEE Journal
on Selected Areas in Communications, 17(8),
1426-1438.
Prasant, M., Jian, L., & Chao, G. (2003, March).
QoS in adhoc wireless network. IEEE Wireless
Communications Magazine, 44-52.
Royer, E.M., & Toh, C.H. (1999, April). A review
of current routing protocols for ad hoc mobile
wireless networks. IEEE Personal Communica-
tions, 6(2), 46-55.
Shen, S.T., & Chen, J.H. (2001). A novel delay-
oriented shortest path routing protocol for mobile
ad hoc networks. In Proceedings of the IEEE
ICC.
Singh, S., & Raghavendra, C.S. (1998, July).
PAMAS-power aware multi-access protocol with
signaling for ad hoc networks. ACM Computer
Communication Review.
Singh, S. Woo, M., & Raghavendra, C.S. (1998a,
October). Power-aware routing in mobile ad-hoc
networks. In Proceedings of the ACM/IEEE
International Conference on Mobile Computing
and Networking, (pp. 181-190).
Singh, S., Woo, M., & Raghavendra, C.S. (1998b,
October). Power-aware routing in mobile adhoc
networks. In Proceedings of Mobicom 98 Confer-
ence, Dallas.
Sivakumar, R., Sinha, P., & Bharghavan, V. (1999,
August). CEDAR: A core extraction distributed ad
hoc routing algorithm. IEEE Journal on Selected
Areas in Communication, 17(8), 1454-65.
Sun, H., & Hughes. (2003). Adaptive QoS routing
based on prediction of local performance in ad hoc
networks. In Proceedings of the IEEE WNCN.
Toh, C.K. (2002, June). Maximum battery life
routing to support ubiquitous mobile computing
in wireless ad hoc networks. IEEE Communica-
tion Magazine, 2-11.
This work was previously published in International Journal of Business Data Communications and Networking, Vol. 4, Issue
1., edited by J. Gutierrez, pp. 52-63, copyright 2008 by IGI Publishing (an imprint of IGI Global).