Content uploaded by Pradeep Gupta
Author content
All content in this area was uploaded by Pradeep Gupta on Aug 31, 2015
Content may be subject to copyright.
ORIGINAL RESEARCH
Performance evaluation of AOMDV routing algorithm with local
repair for wireless mesh networks
Uday Singh Kushwaha
1
•
P. K. Gupta
1
•
S. P. Ghrera
1
Received: 5 August 2014 / Accepted: 3 June 2015
Ó CSI Publications 2015
Abstract Recently, wireless mesh networks (WMNs)
technology has gained a lot of attention in network rese arch
community. WMN is a multi-hop wireless access network
where nodes can act both as a router as well as a host. One of
the factors that influence the performance of WMNs is the
underlying routing protocol used. Thus, numbers of differ-
ent multipath routing protocols are proposed in recent years.
This paper addresses the requirement of local repair under
high speed movement scene of wireless mesh networ k, of
route breakage in multipat h routing in wireless mesh net-
works. In this context we have proposed a local repair based
AOMDV protocol, to maximize the benefit of proposed
algorithm to achieve better performance metrics in both
packet loss ratio and routing overhead and packet delivery
ratio in high speed mobility networks. Various simulations
with varying nodes has been conducted to validate the
improvement of the proposed algorithm against the original
AOMDV. Performance comparison between AOMDV and
AOMDVLR has been analysed in detail, showing that pro-
posed AOMDVLR has achieved better outcome in certain
scenarios.
Keywords Multipath AOMDV AOMDVLR WMN
Local repair NS-2
1 Introduction
Wireless mesh networking (WMN) is a promising com-
munication paradigm and consist of mesh client s (MCs)
and mesh routers (MRs) as shown in Fig. 1. Mesh networ ks
are planned/unplanned multi-hop wireless networks [1].
Mesh routers generally have minimal mobility in a mesh
network and form the backbone of WMNs. This type of
network is very attractive in developing count ries or in the
sparsely populated rural areas where infrastructure is either
non-existent or prohibitively expensive. Due to this
intrinsic property, WMNs have become the focus of
research to increase the coverage range with low cost and
easy deployment [2].
Routing in mesh networks is challenging, since the radio
environment is hostile and unstable, and limited by inter-
ference, with new performance issues. So, many routing
protocols have been proposed in literature by considering
the characteristics of WMN. Here, ad hoc routing algo-
rithms are used due to some similarities between them and
can be categorized in two different category: proactive and
reactive algorithms [3]. In proactive algo rithms the routers
conserve consistent, up-to-date routing information to each
node in the network without considering of whether these
routes are needed or not. Routing tables are updated every
time when the topology changes [4].
Reactive routing protocols are also called as on-demand
approach, In this algorithm, when any new packets needs to
be delivered to the destination and there is no valid route to
carry out this delivery, a new route is to be discovered [5] and
if the route is no longer in use, the routing table will not be
maintained. This type of protocols has low routing overhead
and power consumption, but relatively high route latency [4].
These reactive and proactive protocols are divided into sin-
gle path and multipath routing. The main body of routing
& P. K. Gupta
pradeep1976@yahoo.com
Uday Singh Kushwaha
uday.jptc@gmail.com
S. P. Ghrera
spghrera@rediffmail.com
1
Department of Computer Science and Engineering, Jaypee
University of Information Technology, Solan,
Waknaghat 173234, HP, India
123
CSIT
DOI 10.1007/s40012-015-0065-9
protocol proposals regards single path routing, i.e., for each
source–destination pair a single (shortest) path is discovered
and used for data transmission [6]. Dynamic sourc e routing
(DSR) [5, 7], Ad hoc On-demand Distance Vector (AODV)
[5, 8] are best known single path routing protocol. An
alternative is multi-path routing [9–11] in which mul tiple
paths are used, thereby offering more opportunities for reg-
ulating the traffic over the network. Multi-path routing
enhances single-path routing mainly in two directions: (i) to
have backup paths available in case of path failures and (ii) to
spread traffic to increase the effective bandwidth so that
packet delivery ratio is good. AODV-BR [12] and AOMDV
[13, 14] are best known multipath routing protocol.
As the mobility of nodes increases the network topolo-
gies are changed dynamically. Because of these regular
changes of the network topologies, maintenance and for-
mation of the network is to be difficult. Several multipath
algorithms have been proposed for this purpose that
maintains multiple alternate connections towards the des-
tination along with a primary path with minimum hop
count. When the primary route fails and no alternate path
available towards the destination, rote discovery is started
from source. In WMNs link failure occurs frequently since
nodes in the WMNs are either fixed or mobile. Due to the
high frequency of link failure there should be a mechani sm
that can repair failed link locally or find another path in its
own level instead of sending REER to source to reduce
latency for route recovery.
In this context this paper proposes a Ad hoc on demand
multipath distance vector routing with local repair based
(AOMDVLR) to reducing the routing overhead and packet
loss and increasing packet delivery ratio. AOMDVLR is
applied by the failure sensed node to find new route from
failure sensed no de to destination or next existing node in
the path. Once new paths are available, it starts transmis-
sion from one of the best available with minimum number
of hop count. Since recoveries of routes are done by
intermediate nodes to next available alternate path or
destination, it reduces control overhead and packet loss but
increases packet delivery ratio.
The rest of the paper is ordered as follows: in Sect. 2,
present related work. Section 3 presents proposes AOMDV
with local repair method i.e. AOMDVLR. Section 4 pre-
sents Performance eval uation and results. Section 5 pre-
sents the conclusion and plans for future work.
2 Related work
2.1 Ad hoc on-demand distance vector (AODV)
Ad hoc on demand distance vector is a very popular single
path reactive routing protocol. Routes are set up on demand
basis, and maintain only active routes. This reduces the
routing overhead, but during the on demand route setup it
introduces some initial latency. AODV is based on a simple
request–reply mechanism for the discovery of routes.
AODV protocol mainly involves three packets. They are
route request (RREQ), route reply (RREP) and route error
(RERR). The route request (RREQ) is used for the estab-
lishment of route from source to destination. The route
reply (RREP) is sent by the destination to the source after
the establishment of route. The route error (RERR) is sent
by intermediate node or destination in following
Fig. 1 Infrastructure/backbone
WMNs [17]
CSIT
123
conditions. (i) When there is no path available to the des-
tination (ii) When there is link break in the valid path to the
destination. Source node first broadcast the RREQ packet
by keeping the source and destination IP address in the
RREQ packet. These packets will be received by the
intermediate nodes and these nodes will check whether
there is a suitable route to the destination node or not. If it
has, then the RREQ is further rebroadcasted, otherwise
REER is sent to the source node. The duplicates of route
request packets is discarded by route request id present in
the RREQ packets. When the destination gets the RREQ
packets from different paths, it judge only one path as a
valid path i.e. the path along which it receives the route
request first and other paths are discarded. Destination
replies the RREP packet back to the source with the path
details from source to destination. The main benefit of
AODV is it does not need any centralized scheme to handle
routing process and reacts quickly when topological
changes. But, AODV does not utilize any control or
avoidance mechanism for reducing packet loss. Also, as the
numbers of connection are increased the delivery ratio of
AODV drops dramatically [5, 8, 15].
2.2 The dynamic source routing (DSR)
The DSR protocol is an on-demand reactive routing pro-
tocol based on source routing. In the source routing pro-
cedure, source node determines the accurate sequence of
nodes through which a packet is propagated. In DSR, every
mobile node to maintain a route cache where it caches
source routes that it has learned. Route discovery and route
maintenance are major parts of the DSR protocol. When
any node in the network wants to send a packet to any other
node, it first checks its route cache for a valid route from
source to the destination. If it exists, then the source node
uses this entry to transmit the packet and also attach its
source address on the packet. If it is no t exist in the cache
or the entry in cache is expired (because of long time idle),
the source broadcasts a RREQ packet to all of its neigh-
bours asking for a valid path to the destination. The source
waits till the valid route is discovered towards the desti-
nation. During waiting time, the source can do other tasks
such as sending/forwarding other packets. As the RREQ
packets arrive at intermediate node, it checks from its
caches or from its neighbors whether the required desti-
nation is known or not. If the required route information is
known, it sends back a RREP packet to the source node
otherwise it broadcasts the same route request packet to
their neighbours. When the route is discovered towards the
destination, the data packets are transmitted by the source
node via discovered route. Also, an entry is inserted in the
cache for the future use. The source node also maintains
the age information for each entry to know whether the
cache is fresh enough or not. When a data packet is
received by intermediate nodes, it first checks whether the
packet is intended for itself or not. If it is intended for itself
(i.e. the intermediate node is the destination), the packet
path is attached on the data packet. Since in WMNs any
link might fail anytime route maintenance process, con-
stantly monitors and notifies the nodes if there is any
failure in the exis ting path. Consequently, the node changes
the entries of their route cache otherwise the same is for-
warded using the path attached on the data packet. In
addition, route maintenance process, constantly observe
and notifies the nodes for any change in the entries of their
route cache [5, 7].
2.3 Ad hoc on-demand distance vector—backup
routing (AODV-BR)
AODV-BR protocol is an extensive version of AODV
protocol with a reserve route. AODV-BR route construc-
tion process is based on AODV protocol by flooding of
route request (RREQ) packet. When intermediate nodes get
a RREQ packet from its neighbor, it records the preceding
hop information and the source node information and then
further broadcast the packet to its neighbors or if a route is
known to the desired destination it sends a RREP packet to
the source. The destination replies through a RREP packet
via selected route when it receives the first RREQ or later
RREQs after examine a better route with less number of
hops. It maintains two routing table primary and alternate
routing table. The primary path is established during the
RREQ stage and the alternate paths are established during
the RREP stage. When a node that is not part of the pri-
mary route sensed a RREP packet that is not directed to it,
records the sending neighbor as the next hop to the desti-
nation in its alternate routing table. In this way a node can
receive several RREPs for a single route and select a route
with minimum number of hops count among them and
insert it into the alternate route table. When RREP reaches
at the source of the route, a primary route between source
and destination is constructed. All nodes that have an
alternate route to the destination in their alternate route
table configure a fish bone network [16].
2.4 Ad hoc on-demand multipath distance vector
routing (AOMDV)
This protocol is an extension to the AODV protocol that
compute multiple loop-free and link disjoint paths. The
routing entries for every destination include a list of the
next hops along with the corresponding hop count. All the
next hops contain the identical sequence numb er. This
helps to keep track of a route. Maximum number of hop
count is preserved for each destination by the advertising
CSIT
123
hop count, and defined as the maximum hop count for all
the paths. This hop count is used for sending route adver-
tisements to the destination node. An alternate path to the
destination is defined by duplicate route advertisement
received by a node. Loop freedom is guaranteed for a node
by accepting alternat e paths to destination if it has a fewer
hop count than the advertised hop count for that destina-
tion. Since the maximum hop count is used, therefore the
advertised hop count does not change for the identical
sequence number. The next hop lists as well as the
advertised hop count are reinitialized when a route adver-
tisement is rece ived for a destination with a larger sequence
number. AOMDV may be used to find node disjoint or link
disjoint routes. Each node do not immediately discard
duplicate RREQs to find node disjoint routes. Each RREQs
arriving via a dissimilar neighbour of the source node
defines a node disjoint path. This is because nodes cannot
broadcast replica of RREQs, so any multiple RREQs
arriving at any intermediate nodes via a different neigh-
bour’s node of the source could not have traversed the
identical node. In an attempt to obtain multiple li nk disjoint
routes, the destination replies to replica of RREQs arriving
by unique neighbours. The RREP follows the reverse paths
after the first hop, which is node disjoint and thus link
disjoint. The route of each RREP may overlap at an
intermediate node, but each RREP takes a different reverse
path to the source to make sure link disjointness. Advan-
tage of AOMDV is that it permits to intermediate nodes to
reply to RREQs, while still select disjoint paths. AOMDV
maintains multiple entries towards the destinations; one
entry is selected as primary path and other as an alternative
path. When primary route get fail, alternate path is selected
with minimum number of hops. But, AOMDV has addi-
tional message overheads during route discovery due to
increased flooding and since it is a multipath routing pro-
tocol therefore destination replies to the multiple RREQs
those results are in longer overhead [12, 13].
3 Proposed work
In this paper we proposed a local repair based AOMDV for
the purpose of fault tolerant in WMN s. This protocol works
same as existing AOMDV protocol except route mainte-
nance. In the AOMDV protocol, when no alter nate path is
available then a RERR message is sent to the source node
for the route rediscovery from source to destination. The
proposed AOMDVLR uses different type of control mes-
sages such as RREQ, RREP, HELLO and REER. RREQ is
used to establish forward path from source to destination
and RREP is used to establish backward path destination to
source. HELLO message are used to communicate node
each other. Once the routes has been established the source
node selects a path minimum number of hop count as a
primary path and start transmission via selected primary
path. When the primary path goes down and fails as shown
in Fig. 2, the alternate path is selected with minimum
number of hop count from routing table as shown in
Table 1. The selected alternate path is used for the further
transmission of data packets but when there is no alternate
path available as shown in Fig. 2 it start local repair.
The failure sensed node first check that is has any other
valid alternate path to the destination in its routing table, if
it is found then this selected alternate path with minimum
number of hop count is used for further transmission
otherwise it increment destination sequence number by one
and set TTL in RREQ packet and broadcast it to their old
neighbour and newly came neighbour as shown in Fig. 3
and wait for TTL.
The RREQ receiving node is a newly came node then it
first check freshness of RREQ by destination sequence, if it
fresh RREQ it reply with RREP to their precursor node and
forward this RREQ to next nodes. If the RREQ receiving
node is a old node that have valid path toward the desti-
nation, reply from its cache. If RREP is received within
TTL, it is recorded in the routing table and a new path is
selected with minimum number of hop count as shown in
Fig. 2. Otherwise it generates a RRER message and sends
it to their precursor as shown in Fig. 4.
This precursor node in turn start local repair if it is not a
source node otherwise it will start route rediscovery. A
summarised flowchart for local repair in AOMDVLR is
shown in Fig. 5.
Fig. 2 Node or link break in the active route
CSIT
123
4 Performance evaluation and results
To evaluate the performance improvements of proposed
algorithm we have comp ared the simulation results of
thread protocol AOMDV with and without applying our
scheme. Here, we termed the AOMDV protocol as
AOMDV-LR (AOMDV with local repair) applied on our
algorithm. NS-2 simulator is used to simulate and evaluate
the performance of the AOMDVLR comparing to the
original AOMDV. We have used Red Hat Linux environ-
ment with version NS-2.34 of network simulator. The radio
model uses in the simulation has a nominal bit-rate of
2 Mb/s and a nominal radio range of 250 m. The error-free
wireless channel model is selected in the simulations in
order to isolate the effects of node mobility. The simula-
tions with varying nodes has been performed in a square
field of dimensions 2000 m 9 2000 m. The nodes are
initially placed uniformly at random in the field. The ran-
dom waypoint mobil ity model has been used to simulate
node movements. Pause time is always set to zero. Traffic
pattern consists of several CBR/UDP connections between
randomly chosen source destination pairs. Data packets
have a fixed size of 512 bytes in all the experiments, and
the simulation time is 1000 s. Each data point in the plots is
an average of five such runs with different number of
nodes. Identical traffic and mobility scenarios are used
across all protocol variations. All the summarised simula-
tion parameter is given as in Table 2 .
In Multipath traffic environment with node mobility
following metric are chosen to compare the performance of
routing protocol. The sample screen shot of a scenario of
25 mobile nodes is shown in the Fig. 6.
4.1 Routing overhead
It is the total number of control or routing packets gener-
ated by rout ing protocol during the simulation. The entire
packets sent or forwarded at network layer is consider
routing overhead.
RO ¼
X
n
i¼1
Cs
i
þ
X
n
i¼1
Cf
i
where RO is routing overhead, Cs
i
is control packet sent
and Cf
i
is control packet forwarded.
4.2 Packet loss ratio
Packet loss ratio is the difference betwee n number of data
packet sent and data packet receive in the network.
PLR ¼
P
n
i¼1
Ds
i
P
n
i¼1
Dr
i
P
n
i¼1
Ds
i
100
where PLR is packet loss ratio, Ds
i
is data packet sent and
Dr
i
is data packet received.
4.3 Packet delivery ratio
Packet delivery ratio (PDR) is defined as the percentage of
the ratio between the number of received packets at des-
tination and the number of packets sent by sources. This
performance evaluation parameter measures reliability,
effectiveness and efficiency of a protocol.
Fig. 4 Source restart route discovery after receiving RERR
Fig. 3 Local repair by intermediate route
Table 1 Routing table for Node B
Destination Dest. seq. no. Next hop Hop count Priority notice
D 2 C 2 Active route
D 2 K 3 Sub route
CSIT
123
PDR ¼
P
n
i¼1
Dr
i
P
n
i¼1
Ds
i
100
where PDR is packet delivery ratio, Ds
i
is data packet sent
and Dr
i
is data packet received.
5 Result
In the simulated scena rio the number of CBR/UDP con-
nections is varied while the mean speed is fixed at 5 m/s
and the offered load at 160 kb/s. For a constant mean node
speed and constant offered load, increasing the number of
connections will spread the same amount of traffic among
several connections. This will stress the protocol, as it
requires a routing prot ocol to maintain routes between
more numbers of source–destination pairs. Moreover, each
route discovery will become more expensive because of the
smaller amount of traffic over each connection. Figure 7
demonstrates how protocol routing overhead varies with
varying number of nodes. The general trend observed from
the figure, AOMDVLR outperform AOMDV. This can also
be observed that the routing overhead of AOMDVLR is
lower than AOMDV for all varied connection. As the
Fig. 5 Local repair flowchart
for each node
Table 2 Simulation parameters for AOMDV-LR and AOMDV
S.No. Parameters Values
1 Area size 2000 m 9 2000 m
2 Number of nodes 5–25
3 Node mobility speed 0.9–1.1 v
4 Propagation range 250 m
5 Mobility model Random way point
6 Data rate 5 Kbps
7 Simulation time 1000 s
8 Pause time 0
9 No. of experiments Five times
CSIT
123
number of nodes increases the performance gain by local
repair becomes more significant. This shows that the
AOMDVLR performs better as compared to AOMDV in
all possible number of nodes.
Now we can analyze the Fig. 8. It shows that, packet
loss ratio for AOMDVLR decreases as the number of nodes
increased. It can be concluded that the packet loss ratio of
AOMDV-LR is comparatively better than AOMDV which
is highly real time requirement.
It is maximised the Packet Delivery ratio of proposed
routing protocol due to local route repair of AOMDV.
From Fig. 9 it is observed that, AOMDVLR performance
Fig. 6 Sample simulation scenario with 25 nodes
0
5000
10000
15000
20000
25000
30000
35000
40000
45000
5 10152025
Routing overhead (packets)
Number of nodes
AOMDV
AOMDVLR
Fig. 7 Routing overhead in varied connections
0
1
2
3
4
5
6
7
510152025
AOMDV
AOMDVLR
Fig. 8 Packet loss ratio in varied connections
92.5
93
93.5
94
94.5
95
95.5
96
5 10152025
PDR %
Number of nodes
AOMDV
AOMDVLR
Fig. 9 Packet delivery ration in varied connections
CSIT
123
in terms of PDR is better than AOMDV in different
connections.
6 Summary and discussion
From all the above discussed scenarios, it can be find that
the AOMDVLR protocol can successfully calculate route
and forward data packets in wireless mesh networks sce-
narios better than the original AOMDV in the aspects of
RO, PDR, and PLR.
7 Conclusion and future work
In this paper, the performance of original AOMDV and the
proposed AOMDVLR, which includes an additional
mechanism of local repair, have been analysed. NS-2
simulator has been used to evaluate the performance of the
proposed protocol, in comparison with the original
AOMDV. Using a more active route discovery mechanism,
the proposed AOMDVLR performs better than the original
AOMDV in the aspects of RO, PDR, and PLR. For future
work, the buffer may be used in the intermediates node to
decrease the packet loss value with local repair needs to be
analysed and carefully selected, as well as hardware test-
bed experiments will also be implemented for more
extensive and thorough performance eval uation.
References
1. Akyildiz IF, Wang X, Wang W (2005) Wireless mesh networks: a
survey. Elsevier Comput Netw 47(4):445–487
2. Bicket J, Aguayo D, Biswas S, Morris R (2005) Architecture and
evaluation of an unplanned 802.11b mesh network. In: Proceed-
ings of 11th international conference on mobile computing and
networking, vol. 1, New York, pp 31–42
3. Abolhasan M, Wysocki T, Dutkiewicz E (2004) A review of
routing protocols for mobile ad hoc networks. Ad Hoc Netw 2:1–22
4. Zakaria A, Mohamad H, Ramli N, Ismail M (2013) Performance
evaluation of routing protocols in wireless mesh network. In:
Advanced communication technology (ICACT), 2013 15th
international conference, IEEE, pp 1111–1115
5. Tomar GS, Sharma T, Bhattacharyya D, Kim T-H (2011) Per-
formance comparision of AODV, DSR and DSDV under various
network conditions: a survey. In: Ubiquitous computing and
multimedia applications (UCMA), 2011 international conference,
IEEE, pp 3–7
6. Campista MEM, Passos DG (2008) Routing metrics and proto-
cols for wireless mesh networks. IEEE Netw Mag 22:6–12
7. Johnson D, Hu Y, Maltz D (2007) The dynamic source routing
protocol (DSR) for mobile ad hoc networks for IPv4. RFC4728,
2–100
8. Perkins C, Belding-Royer E, Das S (2003) RFC 3561: ad hoc on-
demand distance vector (AODV) routing. http://www.ietf.org/rfc/
rfc3561.txt
9. Rong B, Yi Q, Lu K, Hu RQ, Kadoch M (2010) Multipath routing
over wireless mesh networks for multiple description video
transmission. IEEE J Sel Areas Commun 28(3):321–331
10. Mao S, Hou YT, Sherali HD, Midkiff SF (2008) Multimedia-
centric routing for multiple description video in wireless mesh
networks. IEEE Netw, Spec Issue Wirel Mesh Netw: Appl, Arch
Protoc 22:19–24
11. Li D, Zhang Q, Chuah CN, Ben Yoo SJ (2006) Multi-source
multi-path video streaming over wireless mesh networks. In:
Circuits and systems, 2006 international symposium, IEEE,
pp 698–701
12. Lee S-J, Mario G (2000) AODV-BR: backup routing in ad hoc
networks. In: Wireless communications and networking confer-
ence, WCNC IEEE, vol. 3, pp. 1311–1316
13. Marina MK, Das SR (2006) Ad Hoc on-demand multipath dis-
tance vector routing. Wirel Commun Mob Comput 6(7):969–988
14. Yuan Y, Chen H, Jia M (2005) An optimized ad-hoc on-demand
multipath distance vector (AOMDV) routing protocol. In: Com-
munications, 2005 Asia-Pacific conference, IEEE, pp 569–573
15. Chennikara-Varghese J, Chen W, et al. (2006) Survey of routing
protocols for inter-vehicle communications, In: IEEE
16. Lee S-J, Gerla M (2000) AODV-BR: backup routing in ad hoc
networks. In: Wireless communications and networking confer-
ence. WCNC IEEE, vol. 3, pp 1311–1316
17. Akyildiz IF, Wang X, Wang W (2005) Wireless mesh networks: a
survey. Comput Netw 47(4):445–487
CSIT
123