AN ADAPTIVE APPROACH FOR ROUTE

OPTIMIZATION IN MANET
Reshma P.R
ME Communication Systems

Ms. S Bharathi
LECT. /ECE DEPARTMENT

M.Kumarasamy college of engineering

M.Kumarasamy college of engineering

Karur,India

Karur,India

[email protected]

[email protected]

ABSTRACT
Geographic routing has been widely hailed as the most
promising approach to generally scalable wireless routing. It
has been a big challenge to develop a routing protocol that can
meet different application needs and optimize routing paths
according to the topology changes in mobile ad hoc networks.
However, there is a lack of holistic design for geographic
routing to be more efficient and robust in a varying
environment. Imprecise information about
local and
destination position can lead to inefficient geographic
forwarding. The use of proactive beaconing schemes to
distribute local positions introduces high overhead when there
is no traffic and cannot capture the topology changes under
high mobility. In this work, two self-adaptive on-demand
geographic routing schemes are proposed which build efficient
paths based on the need of user applications and adapt to
various scenarios to provide efficient and reliable routing. Ondemand routing mechanism in both protocols reduces control
overhead compared to the proactive schemes which are
normally adopted in current geographic routing protocols. The
route optimization scheme adapts the routing path according
to both topology changes and actual data traffic
requirements.The simulation studies demonstrate that the
proposed routing protocols are more robust and outperform
the existing geographic routing protocol and conventional ondemand routing protocols under various conditions including
different mobilities, node densities and traffic loads.
Specifically, the proposed protocols could reduce the packet
delivery latency up to 80 percent. Both routing protocols are
able to achieve about 98 percent packet delivery ratios. They
avoid unnecessary control overhead and have very low
forwarding overhead and transmission delay in all test
scenarios.
GENERAL TERMS
Routing, Proactive and reactive routing, Beacons, Route
request and reply, Simulation
INDEX TERMS
Back Off Period, Beacons, Control Overhead, Geographic
Routing, Local Topology, On-Demand Routing, Optimization,
Recovery Schemes, Route Adaptation, Self-Adaptive Schemes

1. INTRODUCTION
In a mobile ad hoc network (MANET), wireless devices
could self configure and form a network with a virtual
topology. The topology may experience many changes
rapidly and sometimes it will not able to be predicted. Since
all nodes are in mobile nature the design of routing
protocols are much more challenging than that routing
protocols can be categorized as proactive [19], [11], reactive
[12], [13], [14], and hybrid [8], [9], [10]. The proactive
protocol maintain the routing information at every time in
active State, while the reactive ones create and maintain the
routes on the basis of demand. The hybrid protocols
combine the advantage of reactive and proactive
approaches.
The proactive protocols experience high control
overhead when there is no traffic, but for reactive protocols,
the network- range is restricted.it limits their scalability,
and then arises a need for a path from source to destination
prior to the packet transmission. It incurs a large
transmission delay. In recent years, geographic unicast [19],
[20], [15], [5] and multicast [12], [13], [14] routing have
drawn a lot of importance. The assumption is that mobile
nodes are aware of their own positions through global
positioning service or other localization schemes [12], [13]
and a source can obtain the destinations position through
some kind of location service [17], [4]. In geographic
protocols, an neighbor node which acts acts as intermediate
node makes packet forwarding decisions based on its
knowledge of the neighbours positions and the
destinations position inserted in the packet header by the
source. The packets are transmitted greedily to the
neighbour that allows the packet forwarding to make the
greatest geographic progress toward the destination. When
no such a neighbour exists, perimeter forwarding [19], [20]
is used to recover from the local void. In this paper two selfadaptive on-demand geographic routing protocols are
introduced which can provide transmission paths based on
the need of applications. The features of the two protocols
are

They reduce control overhead

The topological information is distributeed on the traffic

demand.
A flexible position distribution mechannism is obtained
by using these protocols
t
change
The forwarding nodes are aware of the topology
in a timely manner and thus more effiicient routing is
achieved.
The optimization schemes are make rouuting paths mor
adaptive to the change of topology and traffic
t
These protocols are robust to the position inaccuracy.
Each node set and adapt the protoocol parameters
independently based on the environmennt change and its
own condition.

The two protocols adopt different schhemes to obtain

information based on topology. First protocol
p
purely
depends on one- hop topology informationn, and the other
protocol is assumed as a hybrid scheme which
w
combines
geographic and topology techniques for more efficient
routing. This scheme avoids the performancee degradation of
conventional geographic routing schemess and it takes
advantage of geographic information to findd each next-hop
thus significantly reducing the overhead andd delay incurred
by network-range search of end-to-end path in conventional
topology-based on-demand routing.
w
include:
To summarize ,the contributions in this work

2.

to Analyse the effect of topology changges and updating

the geographic routing information;
p
which
two geographic routing protocols are proposed
are capable of obtaining and maintaaining topology
information on an adaptive approach;
route optimization schemes are introducced which helps
in adaption of path topology change andd traffic demand;
Adapting parameter are set in both prottocols according
to different conditions,
The destination position inaccuracy is avvoided
Analysing the performance of Qos parameters
p
with
moving speed
SOGR SELF ADAPTIVE
O
ON
DEMAND
GEOGRAPHIC ROUTING PROTOCOLS

Two protocols are explained here. It begins with an

assumption that every node is aware of its geographic
position ie its own position through Gloobal positioning
service.the source can obtain the destinaations position
through some kind of location service. . Inn this paper, f is
the current forwarding node, d is the destiination node, n
denotes one of fs neighbours, posa is
i the position
coordinates of a, and dis(a;b) is the geograaphical distance
between node a and b.

3. SOGR WITH HYBRID REACTIVE MECHANISM

M
The first scheme is SOGR with hybrid
h
reactive
mechanism. The next hop is determined by the
t combination
of geographic and topology based mechanissm reactively.by
using topology-based path searching, a large renge of

topology information can be obtained which helps in finding

a better path. The geographhic routing mechanism helps in
finding the local topologyy. The use of topology-based
routing recovery scheme in SOGR
S
helps to overcome many
shortcomings of geographic routing.
r

Let D be the destinationn and F be source. F will search

for D in one hop limit. If D is in one hop limit then the
message is sent unicastly. If D is not in one hop limit then it
will send route request to all its neighbor nodes.At the
destination SINR thresholdd is set initially. When route
request reaches the destinatioon through different nodes then
destination compares its thrreshold with SINR of various
routes. The route having highher SINR represent high energy
route. The reply is sent throuugh the highest energy route.
3.1 Geography-Based Greeedy Forwarding
Greedy forwarding is one of the forwarding mechanism
used by scheme 1.F willl forward a packet close to
destination D than itself greedily.if
g
there is no next hop
then F will buffers the packket and then broadcast a request
message.the request messsage will be in the form
REQ(D,posd,posf,h).Initiallyy h is set as 1.
If a neighbour node N closer to D than F sends
shortcomings of geographic routing. Back a REPLY, F will
record N as the next-hop to D with the transmission mode
set as greedy and unicast thee packet. If next REPLY arrives
from a node N later, F then updates its next-hop to N only
if N is closer to D than N, or else ignore the reply . REQ
has a small size and a higher probability of being
transmition. Failures in trannsmission can be avoided by a
node repling only if the received signal to noise plus
interference ratio of it received REQ is above a threshold set
higher than the target decodding need.a neighbour N waits
for a back off period upto which
w
a REPLY will not be send
back inorder to avoid collisioon and the pending REPLY will
be canceled if it overhears either a REPLY from another
neighbour closer to D than ittself or the packet sending by F

with the next-hop closer to D than N, indicating that F

received a REPLY already. For avoiding collisions among
replies D respond only after a backoff period and all request
after backoff period are suppressed. The back off period
tnbf should be proportional to dis(N;D) and bounded by the
max value h x Ibf , where Ibf is a protocol parameter, and
the hops h is set to 1 in greedy forwarding. For Each node N
the backoff period is calculated as

TbfN

Ibf

dis F, D
h

dis N, D
R

Where R is the reference transmission range of mobile

nodes. If multiple neighbours have very similar distances to
destination,then their reply messages may have a tendency
to get collide. To avoid this, we make use of a parameter
which is initialized to 1 when F sends out the first request
message to ensure that the nodes nearer to D reply soon. It
is set to a random number ranges between 0 and 1 during
recovery forwarding to avoid reply collisions from
neighbours that are of equal distance to D. F will broadcasts
the first REQ message, if multiple neighbours have similar
closest distance to D and collide in their replying while F
gets a reply from a node that has a larger distance to the D,it
will not be affected since the next-hop found is not the one
closest to D. A node closer to Destination than the current
next hop can send a CORRECT message later to F through
the optimization process If all the reply messages are lost, a
neighbouring node is given a further opportunity of sending
back its REPLY during the recovery forwarding.
TABLE 2
Values Used in SOGR-HR and
SOGR-GRs Adaptive Parameter Settings
Refbf
bf
Dist
[It,min,It,max]
Disbc
[Ibc,min,Ibc,max]

Values
10ms
2ms
300m
[10s,30s]
150m
[5s,15s]

protocol
SOGR-HR
SOGR-HR
SOGR-HR
SOGR-HR
SOGR-GR
SOGR-GR

If F does not receive any reply within 1:5 x h x Ibf , F

will initiate a process for recovery. There are two reasons
for F to fail in getting reply message:
1) The reply messages from all its neighbours are lost.
2) F may not have neighbours closer to D, resulting in a
local void.

Without knowing the local topology, the recovery

schemes [20], [19], [10], [11] based on planar structure
cannot be used to address problems. SOGR-HR uses a
recovery strategy with expanded ring search (which is
normally used in path finding in topology based routing
protocols [13], [12]) to build a more efficient path to recover
from the local void by taking advantage of larger range
topology information.
In a recovery process, F increases its searching limit to
more number of hops. Sometimes the REPLY on the first
try may be lost due to collisions. Whenever a REQ reaches a
one-hop neighbour that is closer to D than F, the neighbour
sends back a REPLY after a back off period according to
tbfn with h = 1. Otherwise, the one-hop neighbour of F
continues broadcasting the REQ to its own one-hop
neighbours. When a second-hop neighbour of F gets this
REQ and is closer to D, it sends a REPLY which follows the
reverse path of the REQ message which is followed from F
to D, with the back off period calculated from tbfn at h = 2.
Different from that in greedy forwarding, the here is set to
a random number between 0 and 1 for both one-hop
neighbours and two-hop neighbours to avoid potential reply
collisions from neighbours that have similar distance to the
destination.
When a REPLY is sent by a two-hop neighbour, the
intermediate nodes record the previous hop of the REPLY
as the next-hop toward D with the transmission mode which
is set as recovery. But when the REPLY is originated from a
neighbour who is one-hop of F, F set to greedy transmission
mode.for avoiding overhead, an intermediate node drops a
REPLY if it already forwarded or overheard a REPLY from
a node closer to D than the current node. F send the data
packet unicastly to the detected next hop with the
corresponding transmission mode. If the route searching
fails with h = 2, F may expand the searching range again by
increasing the value of h until it reaches Maxhops. Instead
of searching for an end-to-end path, the topological
information is used for searching and selection of relay
node(s) toward D. Recovery forwarding is triggered only
when needed and the relay nodes can generally be found
within a small range.

4.

SOGR with Geographic-Based

Reactive Mechanism (SOGR-GR)

SOGR-GR depends only on one-hop neighbours

positions to make greedy and perimeter forwarding like
other geographic routing protocols [20]. However, it adopts
a reactive beaconing mechanism which is adaptive to the
traffic scenarios. The beaconing is triggered only when a
node overhears data traffic from its neighbours for the first
attempt. The beaconing ends if there is no traffic heard for a
period which is predefined. A node which act as forwarding
node broadcasts a request (REQ) message to trigger its
neighbours beaconing when necessary, and the neighbours

will have random back off before broadcastting a beacon to

avoid collision.

routing due to the inaccuraccy in topological information.

Both protocols can handle opptimization mechanism.

With the neighbour position informattion, SOGR-GR

takes over the recovery method as exissting geometric
routing protocols to avoid the need of extraa searching as in
SOGRHR. Like SOGR-HR, the protocol parameters of
mal performance.
SOGR-GR are also set adaptively for optim
To make the beacon sending on demand, evvery node keeps
three time values treq, treqHeard, and tbcc, in which treq
records the time when the latest REQ or data
d
packet was
sent out, treqHeard is the time when the lateest REQ or data
transmission was heard, and tbc saves thee last beaconing
time.

Let S be source and D be destination. At the time of link

breakage source will sends beacons to all neighbor
n
nodes.
The data table will be updated and new rouute is found and
data is transferred through that route.. Beacons are
transmitted upto a maximum of two times. If
I any neighbor
node is not responding to beacons upto two times
t
then those
nodes will be marked as lazy nodes.
4.1 Route Adaptation and Optimization

As the nodes are moving in random

m there will be
frequent topology changes.so route optimiization schemes
helps to adapt for topology changes and
traffic
environment. The topological informatioon is evaluated
before forwarding the packet so as to avvoid forwarding
failure due to outdated information of neighbbor.
The routing path is optimized by inncorporating the
forwarding node and its neighbours to avvoid undesirable

MIZATION FOR THE

5. OPTIM
FOR
RWARDING PATH
Both these algorithms forrms a step by step procedure so
that loop free networks can be obtained. To achieve more
optimal routing during freequent topology changes Fs
neighbours monitor whetheer F makes correct forwarding
decisions and help to improve transmission path
f
a packet to C which
opportunistically. After F forwards
continues the forwarding toward D, a neighbour N
overhears both transmissionns and gets posF , posC, and
posD. A packet forwarded using the recovery mode will
also carry the position of thhe node (say node S) where the
recovery forwarding is origginated, posS. If N determines
that it is a more optimal neext hop than C, it sends to F a
message CORRECT(posN;D
D) asking it to convert its next
hop to node N. Three optiimization cases are considered
here. With mode(A;B;D) reppresenting the forwarding mode
from A to B toward a deestination D. there are certain
criterias for N for sendingg a CORRECT message is as
follows:

Case 2: for mode(F;C;D) greedy method is used. When

node N is closer to D than C is, i.e., dis(N;D) < dis(C;D),
then node N will inform F to set its new next hop to N.
Case 3: mode(F;C;D) is in recovery mode condition.
There are two cases:
a)

the last hop of the recovery mode is F, ie dis(C;D)

< dis(S;D) and if dis(N;D) < dis(C;D),then F
should forward its future packets to N for a better
optimal route.

b) if F is not the last hop of the recovery forwarding,

so dis(S;D)dis(C;D). If dis(N;D) < dis(S;D), then
F forward the packet to N and it undergoes greedy
forwarding.

By this process, much more optimal forward routing

can be obtained. In cases 2 and 3, to avoid that multiple
neighbours detect nonoptimal forwarding simultaneously
and send CORRECT messages to F at the right time. It will
be sent with back off period and that is done for REPLY
message with h = 1. Without a recovery forwarding phase as
for next-hop finding, the parameter is set as a random
number between 0 and 1 to further reduce message
collision. Assume that recovery forwarding starts at Fand if
F sets its next hop C to reach node T .this node is closer to
D when compared to F. F is unaware of the positions of non
neighbouring nodes on the recovery path to T. on the
recovery path a node should notify F with an ERROR
message at the time whenever it detects next hop as
unreachable. T should also notify F if it is no longer closer
to D than F . Then F will initialize a new route searching
process. SOGR-GR and SOGR-HR uses similar schemes for
route optimization.

off interval is Refbf. We restrict the range for searching of

SOGR-HR to two hops ie setting maximum hops (Maxhops)
as two because in most cases nodes closer to the destination
can be found within this range and a larger searching range
will result in a bigger control overhead.
The simulations were run with 300 nodes randomly
distributed in an area of 3000 m x 1500 m. We chose a
rectangular network area to obtain a longer path. The
moving pause time is set to 0 second. Initially the minimum
speed is set to 0 m/s, and the default maximum speed is set
to 20 m/s .IEEE 802.11b was used as the MAC layer
protocol to coordinate medium access and resolve
collisions. Each simulation last till 900 simulation
seconds.traffic flow is sent at 8 Kbps using CBR between a
source and destination pair .The packet length is 512
bytes.30 CBR flows are used in the simulations. A
simulation result was gained by averaging over 20 runs with
different seeds to increase the confidence of the results. The
receiver power is set to 0.01 and transmitter power is set to
0.02. The initial energy is set to 100.

Packet delivery ratio

Case 1: When N (destination) moves into Fs

transmission range, F forwards the packet to N.

1
0.98
0.96
0.94
0.92
0.9
0.88
0.86
0.84
0.82
0.8

10
20
30
40
Maximum moving speed (m/s)

50

(5a)
8
Data packet forwarding
overhead

For SOGR-HR and SOGR-GR the parameter set

initially. all the parameters are adaptive .These parameters
are adjusted at each forwarding and the initial values are not
critical. The parameter Ijitter in SOGR-GR is set to 10ms.
The reference distance threshold Disbc for a beacon update
in SOGR-GR is set to be smaller than the transmission
range. The time-out reference distance Dist for SOGR-HR is
set to be double Disbc so that the time-out periods for
SOGR-HR and SOGR-GR are comparable. The initial back

SOGR-GR

6. SIMULATION OVERVIEW
We implemented SOGR-HR and SOGR-GR with ns2.
the protocols are on-demand and geography-based, for
performance evaluations, we are comparing our on demand
protocols SOGR-HR and SOGR-GR in this paper.

SOGR-HR

SOGR-HR

7
SOGR-GR

6
5
4
3
2
1
0
0

10
20
30
40
Maximum moving speed (m/s)
(5b)

50

110

We analyse the performance of mobility of nodes for

various QOS parameters for both SOGR-HR and SOGRGR. The moving speed adjusted from 0 to 50 m/s. In fig.5a
the delivery ratios of the two protocols decrease quickly as
the moving speed Increases. They Determine the next hop
based on the knowledge of Local topology, and can respond
to the mobility much faster. SOGR-HR and SOGR-GR
maintains a High delivery ratio of around 99 percent even in
a highly Dynamic environment of random nodes. For
SOGR-HR and SOGR-GR, the adaptive parameter Settings
are more flexible .The position information will Intelligently
generate control messages to distribute Topological
information and better tracking mobility.in fig.5b both
SOGR-HR and SOGR-GR generates more control messages
to Capture the topology changes as mobility of node
increases. SOGR-HR generates a slightly higher control
Overhead than SOGR-GR because whenever the next Hop
is not valid, the forwarding node will start searching for a
new route .in SOGR-GR, the forwarding node just Selects
next hop from its neighbor routing Table without making
extra control overhead. In Fig. 5c SOGR-GR has a more
forwarding overhead than SOGR-HR since the perimeter
12

Avg. end to end delay

100

7. SIMULATION RESULTS
7.1 Effect of Moving Speed

90
80

SOGR-HR

70
SOGR-GR

60
50
40
30
20
10
0
0

10

20

30

40

50

Maximum moving speed (m/s)

(5d)
In summary The two OGR protocols are robust to
topology changes under high mobility.they utilizes the
position information more precisely and adaptively in
response to different speed levels of nodes. For more precise
position information, various optimization schemes are
adopted by both SOGR-HR and SOGR-GR.they have much
fewer redundant transmissions and they have low end-toend delay.
8. CONCLUSIONS

Control overhead

10
SOGR-HR

SOGR-HR

6
4
2
0
0

10
20
30
40
Maximum moving speed (m/s)

50

(5c)
forwarding. In SOGR-GR will introduce more packet
forwarding overhead. SOGR-HR can make out more better
routing path Without limiting to one-hop. In Fig. 5d Both
SOGR protocols are able to achieve Small delay with the
use of various path Optimization strategieswhich are
adaptive to track the topology changes in a Timely manner.
SOGR-HR starts a new next-hop search Whenever the next
hop is invalid.hence SOGR-HR has longer delay than
SOGR-GR.

In this work, we propose two protocols which are selfadaptive on-demand geographic routing protocols - SOGRHR and SOGR-GR. The two protocols adapts to different
schemes to obtain local topology information and it is
maintained. SOGR-GR purely depends on one-hop topology
information for forwarding as other geographic routing.The
simulation results demonstrate that our protocols are very
robust in a dynamic mobile network. They can efficiently
adapt to different conditions and can perform better than
other geographic routing protocols that exist. Both proposed
routing protocols could achieve about 98 percent packet
delivery ratios by avoiding unnecessary control overhead.
By using these we can have very low-forwarding overhead
and transmission delay in all test scenarios. Moreover this
paper concentrates on reducing the redundancy to establish
path.
REFERENCES
[1]

[2]

[3]

Y.C. Tseng, S.Y. Ni, and E.Y. Shih, Adaptive Approaches to

Relieving Broadcast Storms in a Wireless Multihop Mobile Ad Hoc
Network, IEEE Trans. Computers, vol. 52, no. 5,pp. 545- 557, May
2003.
Y. Cai, W. Lou, M. Li, and X.-Y. Li, Energy Efficient TargetOriented Scheduling in Directional Sensor Networks, IEEE Trans.
Computers, vol. 58, no. 9, pp. 1259-1274, Sept. 2009.
W. Wu, J. Cao, J. Yang, and M. Raynal, Design and Performance
Evaluation of Efficient Consensus Protocols for Mobile Ad Hoc
Networks, IEEE Trans. Computers, vol. 56, no. 8, pp. 1055-1070,
Aug. 2007.

[4]

[5]

[6]

[7]

Abraham, D. Dolev, and D. Malkhi, LLS: A Locality Aware

Location Service for Mobile Ad Hoc Networks, Proc. Workshop
Discrete Algorithms and Methods for MOBILE Computing and
Comm.(DialM), 2004.
S. Basagni, I. Chlamtac, V. Syrotiuk, and B.A. Woodward, A
Distance Routing Effect Algorithm for Mobility (DREAM),
Proc.ACM/IEEE MobiCom, pp. 76-84, 1998.
L. Blazevic, S. Giordano, and J.-Y.L. Boudec, Self Organized
Terminode Routing, Cluster Computing J., vol. 5, no. 2, pp. 205218, 2002.
H. Fussler, J. Widmer, M. Kasemann, M. Mauve, and Hartenstein,
Beaconless Position-Based Routing for Mobile Ad- Hoc Networks,
Technical Report TR-03-001, Dept. of Math. and Computer Science,
Univ. Mannheim, Germany, 2003.

[8] Z. Haas, M. Pearlman, and P. Samar, Zone Routing Protocol

(ZRP), IETF Internet draft, July 2002.
[9] V. Ramasubramanian, Z.J Haas, and E.G Sirer, SHARP: A
Hybrid Adaptive Routing Protocol for Mobile Ad Hoc
Networks, Proc. ACM MobiHoc, June 2003.
[10] R. Beraldi and R. Baldoni, A Caching Scheme for Routing in
Mobile Ad Hoc Networks and Its Application to ZRP, IEEE
Trans. Computers, vol. 52, no. 8, pp. 1051-1062, Aug. 2003.
[11] P. Jacquet, P. Muhlethaler, and A. Qayyam, Optimized LinkState Routing Protocol, IETF MANET Internet draft, Mar.
2002.
[12] C. Perkins, E.M. Royer, and S. Das, Ad Hoc On-Demand
Distance Vector (AODV) Routing, IETF RFC 3561, July
2003.
[13] D.B. Johnson, Y. Hu, and D.A. Maltz, The Dynamic Source
Routing Protocol (DSR) for Mobile Ad Hoc Networks for
IPv4, IETFRFC4728,Feb2007.