Fifth International Conference on Information Technology: New Generations

Stable and Energy Efficient Routing for Mobile Adhoc Networks

Vinay Rishiwal Ashwani Kush Shekhar Verma

Department of CS & IT, Department of Computer Sceince, Department of Info. Technology,
MJP Rohilkhand University, Kurushetra University, IIIT Allahabad,
Bareilly, UP, India. Kurushetra, India. Uttar Pradesh, India.
[email protected] [email protected] [email protected]

Abstract environments, communications set up in Exhibitions,

Mobile Ad hoc networking is a challenging task due to conferences and meetings, in battle field among soldiers
the lack of resources in the network as well as the to coordinate defense or attack, at airport terminals for
frequent changes in network topology. Although lots of workers to share files etc. Several routing protocols for ad
research has been done on supporting QoS in the Internet hoc networks have been proposed in the literature. In
and other networks, but they are not suitable for mobile most of the routing protocols, major emphasis has been on
Ad hoc networks and still QoS support for such networks finding shortest routes. In this paper a new scheme the
remains an open problem. In this paper, a new scheme has stable routing with power factor (SBNRP) has been
been proposed for achieving QoS in terms of packet suggested which would allow mobile nodes to maintain
delivery, multiple connections, better power management routes to destinations with more stable route selection.
and stable routes. It offers quick adaptation to distributed This scheme responds to link breakages and changes in
processing, dynamic linking, low processing overhead network topology in a timely manner.
and loop freedom at all times. The proposed scheme has Due to the frequent changes in network topology
been incorporated using AODV protocol and by extensive and the lack of the network resources both in the wireless
simulation the performance has been studied, and it is medium and in the mobile nodes, mobile ad hoc
clearly shown that the proposed scheme performs very networking becomes a challenging task. Routing in Ad
well for different network scenarios. hoc networks experiences more link failures than in other
networks. Hence, a routing protocol that supports QoS for
Key Words: Ad hoc network, Routing protocols, AODV, ad hoc networks requires considering the reasons for link
Power, Quality of Service. failure to improve its performance. Link failure stems
from node mobility and lack of the network resources.
Therefore it is essential to capture the aforesaid
1. Introduction and Motivation characteristics to identify the quality of links.
Furthermore, the routing protocols must be adaptive to
An ad hoc wireless network is a collection of cope with the time-varying low-capacity resources. For
mobile devices equipped with interfaces and networking instance, it is possible that a route that was earlier found
capability [1]. Such devices can communicate with to meet certain QoS requirements no longer does so due
another node within their radio range or one that is to the dynamic nature of the topology. In such a case, it is
outside their range by multi hop techniques. An Ad hoc important that the network intelligently adapts the session
network is adaptive in nature and is self organizing. In to its new and changed conditions. Quality of service [11]
this wireless topology may change rapidly and means providing a set of service requirements to the flows
unpredictably. A mobile ad hoc network is also called while routing them through the network. A new scheme
MANET [2]. The main characteristic of MANET strictly has been suggested which combines two basic features to
depends upon both wireless link nature and node mobility achieve QoS; these are stable routing and concept of
features. Basically this includes dynamic topology, battery power. The scheme uses backbone nodes for
bandwidth, energy constraints, security limitations and stable routes and uses power factor to determine active
lack of infrastructure. MANET is viewed as suitable nodes to participate in routing.
systems which can support some specific applications as The rest of the paper is organized as follows:
virtual classrooms, military communications, emergency Section 2 takes a look at the Routing protocols
search and rescue operations, data acquisition in hostile classification, Section 3 analyzes new proposed scheme

978-0-7695-3099-4/08 $25.00 © 2008 IEEE 1029

1028
DOI 10.1109/ITNG.2008.230
(SBNRP), Section 4 describes the simulation environment protocol is divided into three phases. Route Request
and results and Section 5 summarizes the study and the (REQ), Route Repair (REP) and Error Phase (ERR).
status of the work.
S 2
2. Routing Protocol Classifications 3 4
1 D
A routing protocol is needed whenever a packet 6
needs to be handed over via several nodes to arrive at its 5 8 7
12
destination. A routing protocol has to find a route for
packet delivery and make the packet delivered to the 11
correct destination. Routing Protocols have been an active 9
area of research for many years; many protocols have 10
been suggested keeping applications and type of network
in view. Routing Protocols in Ad Hoc Networks can be
classified into two types: Stable node Less stable node

2.1 Table Driven or Proactive Protocols

Figure 1: An Example of routing
In Table Driven routing protocols each node
maintains one or more tables containing routing The proposed scheme is explained with the help
information to every other node in the network. All nodes of an example shown in Figure 1. Assume that the node
keep on updating these tables to maintain latest view of with index 1 is the source while destination is the node
the network. Some of the famous table driven or proactive with index 4. Note that the route discovered using new
protocols are: GSR [3], WRP [4], ZRP [5], STAR [6] etc. scheme routing protocol may not necessarily be the
shortest route between a source destination pair. If the
2.2 On Demand or Reactive Protocols node with index 3 is having power status in critical or
In On Demand routing or reactive protocols, danger zone, then though the shortest path is 1—2—3—4
routes are created as and when required. When a but the more stable path 1—2—5—8—9—10—4 in terms
transmission occurs from source to destination, it invokes of active power status is chosen. This may lead to slight
the route discovery procedure. The route remains valid till delay but improves overall efficiency of the protocol by
destination is achieved or until the route is no longer sending more packets without link break than the state
needed. Some famous on demand routing protocols are: when some node is unable to process route due to
DSR [7], RDMAR [8], AODV [9] etc. The study has inadequate battery power. The process also helps when
been concentrated for Reactive Protocols due to dynamic some intermediate node moves out of the range and link
topology of network. Surveys of routing protocols for ad break occurs, in that case backbone nodes take care of the
hoc networks have been discussed in [2, 10]. process and the route is established again without much
overhead. In Figure 1 if the node with index 8 moves out,
the new established route will be 1—2—5—11—9—10—
3. Proposed Scheme: SBNRP 4. Here the node with index 11 is acting as backbone node
(BN) for the node with index 5 and the node with index 8.
The proposed scheme takes care of on demand Similarly the node with index 12 can be BN for the nodes
routing along with a new concept of backbone nodes with with indices 7, 10 and 4. BN has been selected at one hop
power factor. Many protocols have been discussed using distance from the said node.
concept of power in many existing schemes [12-20]. In all
the schemes discussed under concept of power routing, no 3.1 Route Construction (REQ) Phase
concern has been taken for stable routing or better packet
delivery. All emphasis is on concept of battery power or This scheme can be incorporated with reactive
energy requirement for routing process. routing protocols that build routes on demand via a query
In this paper two different concepts have been and reply procedure. The scheme does not require any
joined together to make an efficient protocol. Major modification to the AODV's RREQ (route request)
concentration is on the routing problem. In the proposed propagation process. In this scheme when a source needs
scheme, the backbone nodes help in reconstruction phase to initiate a data session to a destination but does not have
in fast selection of new routes. Selection of backbone any route information, it searches a route by flooding a
nodes is made upon availability of nodes and battery ROUTE REQUEST (REQ) packet. Each REQ packet has
status. Each route table has an entry for number of a unique identifier so that nodes can detect and drop
backbone nodes attached to it and their battery status. The duplicate packets. An Intermediate node with an active

1029
1030
route (in terms of power and Backbone Nodes), upon MIN_REPAIR_TTL+BN, so the whole process becomes
receiving a no duplicate REQ, records the previous hop invisible to the originating node
and the source node information in its route table i.e. This factor is transmitted to all nodes to select
backward learning. It then broadcasts the packet or sends best available path with maximum power.
back a ROUTE REPLY (REP) packet to the source if it
K K1
has an active route to the destination. The destination
node sends a REP via the selected route when it receives L1 M
the first REQ or subsequent REQs that traversed a better P Q
active route. Nodes monitor the link status of next hops in L Destination
active routes. When a link break in an active route is
detected, an ERR message is used to notify that the loss of A C
link has occurred to its one hop neighbor. Here ERR
message indicates those destinations which are no longer Source
P1 P2
reachable. Taking advantage of the broadcast nature of
wireless communications, a node promiscuously
Link break
overhears packets that are transmitted by their
neighboring nodes. When a node that is not part of the
Figure 2: Local Repair
route overhears a REP packet not directed to itself
transmit by a neighbor (on the primary route), it records
that neighbor as the next hop to the destination in its Figure 2 gives an idea of working of local route repair.
alternate route table. From these packets, a node obtains Initial path from source node ‘Source’ to destination node
alternate path information and makes entries of these ‘Destination’ is shown via solid lines. When link breaks at
backbone nodes (BN) in its route table. If route breaks node C, route repair starts, node C starts searching for
occurs it just starts route construction phase from that new paths, buffering packets from S-A in its buffer. Node
node. The protocol updates list of BNs and their power C invokes Route Request phase for ‘Destination’.
status periodically in the route table.
. Table 1: Active Time Estimations
3.2 Route Error & Maintenance
Node BN Min_T #hops*.5 Power Total
In this scheme data transmits continuously status
TL
through the primary route unless there is a route
disconnection. When a node detects a link break, it L= 3 3 1/2 9 15
performs a one hop data broadcast to its immediate
neighbors. The node specifies in the data header that the M= 4 2 2/2 8.5 14.5
link is disconnected and thus the packet is candidate for
K= 3 1 3/2 8 12
alternate routing. Upon receiving this packet route
maintenance phase starts by selecting alternate path and P= 3 1 2/2 4 8
checking power status.
Q= 3 1 3/ 2 3 07
3.3 Local Repair (ERR Phase) P1= 1 4 1/2 7 12.5

When a link break in an active route occurs, the P2= 2 3 2/2 7 12

node upstream of that break may choose to repair the link
locally if the destination was no farther and there exists K1= 2 1 4 /2 7.5 10.5
BNs that are active. The Time to live (TTL) of the REQ L1= 1 3 2/2 8.0 12
should initially be set to the following value:

TTL = max (MIN_REPAIR_TTL + BN, 0.5 * #hops) + Now backbone nodes are selected and proper selection of
Power status nodes is done based on power factor. Path selected
Where the MIN_REPAIR_TTL is the last known hop becomes [C - L – M – K – Destination], instead of [C –
count to the destination, #hops is the number of hops to L – P –Destination], since the node P is not in active
the sender (originator) of the currently undeliverable state. Even though the route may become longer, but the
packet. BN is number of backbone nodes attached to the selected route path is far more stable and delivers more
said node and Power status is power state of the node at packets. Stability of route depends upon two major
that time. As 0.5* #hops is always less than aspects as: Life time and Power status. The concept has
been explained in Table 1.

1030
1031
 When selection has to be made between nodes
P1 and L at the start of repair phase, selection of node L Packet Delivery Ratio 100 nodes
has the advantage over node P1. Similarly in the selection
between nodes K and K1, node K has higher weight. If 1
any BN has not been on active scale, it is rejected and a

p a ck e t d e li v e ry r ati o
0.95
new node is searched. In addition to power factor, efforts SBNRP
are made to keep the path shortest. This local repair 0.9 AODV
attempts are often invisible to the originating node. DSR
During local repair data packets will be buffered 0.85
TORA
at local originator. If, at the end of the discovery period, 0.8
the repairing node has not received a reply message REP
it proceeds in by transmitting a route error ERR to the 0.75
originating node. On the other hand, if the node receives 0 20 50 100 150 200 250 300 350 400 450 500
one or more route reply REPs during the discovery Pause Time
period, it first compares the hop count of the new route
with the value in the hop count field of the invalid route Figure 3: Packet delivery ratio at diff. Pause time
table entry for that destination. Repairing the link
locally is likely to increase the number of data packets
that are able to be delivered to the destinations, since data
packets will not be dropped as the ERR travels to the packet delivery ratio 100 nodes
originating node. Sending a ERR to the originating node
after locally repairing the link break may allow the 100
originator to find a fresh route to the destination that is 98
better, based on current node positions. However, it does
not require the originating node to rebuild the route, as the 96 SBNRP
ra tio

originator may be done, or nearly done, with the data aodv

94 dsr
session. In AODV, a route is timed out when it is not used
and updated for certain duration of time. The proposed 92
scheme uses the same technique for timing out alternate
routes. 90
1 2 5 10 15 20

4. Simulation and Results speed m/s

Simulation study has been carried out to study Figure 4: Packet delivery ratio at diff. speed
the Performance study of existing different protocols.
Simulation Environment used is NS-2 (network End-to-End Delay
simulator) version NS2.29 to carry out the process. The
simulator NS-2 has been developed by University of Average end-to-end delay is the delay
California and the VINT project [15]. Simulation results experienced by the successfully delivered packets in
have been compared with AODV, DSR and TORA. reaching their destinations. This is a good metric for
Simulation study has been performed for packet delivery comparing protocols and denotes how efficient the
ratio, Throughput and End to End delay evaluations. underlying routing algorithm is, because delay primarily
depends on optimality of path chosen
Packet Delivery Ratio: The fraction of successfully
S
received packets, which survive while finding their 1
destination. This performance measure also determines Average end-to-end Delay =
S
∑ (r
i =1
i − si )
the completeness and correctness of the routing protocol.
If C is total number of flows, f is id, R is packets received
from f and T is transmitted from f, then F can be
determined by
1 C Rf
F= ∑
C f =1 T f

1031
1032
 cache is too large to benefit performance. Often stale
End to End delay - 100 nodes sbnrp routes are chosen and all this leads to more packet falls.
22 dsr AODV is delivering more packets to DSR in most of the
20
aodv cases and has an edge over it. New scheme (SBNRP) is
tora overall best for 100 nodes. It starts with 86% and with
e n d to e n d d el ay

18
increasing pause time gets stable and delivers more than
16 95% packets.
14 Figure 4 shows the simulation results with speed as a
function. AODV and DSR have performed better at all
12
speeds. DSR cache performance has suffered a bit at
10 higher speed in denser medium. The reason is that
0 20 50 100 150 200 250 300 400 500 keeping cache for such a large network demand more
pause time storage and in turn slows packet delivery rate. DSR is
able to deliver more than 94% packets all the time.
AODV improves in denser mediums as it is able to
Figure 5: End to end delay at diff. pause time support more packets. It overpowers DSR at high speed of
15 to 20 meters per second and trend is true even at higher
Throughput speeds. New scheme has been the best in dense mediums,
It is defined as rate of successfully transmitted showing almost same performance at all speeds. In case
data per second in the network during the simulation. of New scheme delivery ratio was nearing 98% even at
Throughput is calculated such that, it is the sum of higher speeds 10 to 20 and more. Proposed scheme
successfully delivered payload sizes of data packets outperforms all other schemes. This proves New scheme
within the period, which starts when a source opens a performs better in denser medium, as more backbone
communication port to a remote destination port, and nodes are available for route selection and also more
which ends when the simulation stops. Average nodes are available with better power status.
throughput can be calculated by dividing total number of
bytes received by the total end-to-end delay. End to end delay has been explained in Figure 5.
Here it is clear that DSR has more delays than AODV.
The protocol proposed has higher delays. While DSR uses
Throughput 50 nodes SBNRP source routing, it gains so much information by the source
AODV that it will learn routes to many destinations than a
35000 DSR distance vector protocol like AODV or New. This means
30000 tora while DSR already has a route for a certain destination,
th r o u g h p u t b y te s / s

25000 New would have to send a specific request for that

20000 destination. The packets would in the meanwhile stay in
15000 the buffer until a valid route is found. This takes some
10000 time and will therefore increases the average delay. The
5000 delay for New is more and the reason is that it spends
0 more time in calculation of stable route. New does deliver
0 0.5 1 1.5 2 2.5 3 3.5 even those packets, which may have been dropped in
speed m/s AODV as it has better recovery mechanism and local
repair system for faster recovery. All this process
Figure 6: Throughput at different speeds increases delay but not at the cost of efficiency.
Throughput in bytes per second has been
In simulation study 100 nodes were taken in a random calculated and speed has been varied from 0 to 3.5 meter
scenario of size 1000 × 1000. Two parameters have been per second. Figure 6 shows the graphical representation.
takes as Pause time and speed. The study has been TORA can spare less bandwidth to data transmission. It
conducted at different pause times. Pause time of 0 means has almost steady throughput. DSR, AODV and New
maximum mobility and 500 is minimum mobility. The have an increase in throughput. The throughput increase
sources connected are 25-34 using TCP connection. can be further explained by TCP behavior, such that,
Figure 3 represents the results. TORA is not delivering when ACK is not received back, TCP source retransmits
more than 84% in denser mediums. It is unable to perform data. With increased mobility, source may distribute
better in higher congestion zones. AODV outperforms several identical packets to different intermediate nodes,
DSR in congested medium. DSR drops significant packets because of route changes At 1.5 m/s speed, AODV
at high speed; the reason can be use of source routing and protocol also shows an increasing behavior with the
aggressive use of cache. With higher loads the extent of increased network speed. This is due to increased

1032
1033
 triggered replies and requests. New scheme performance 10. Josh Broch, David A.Maltz and Jorjeta Jetcheva, “A
is driven using concept of AODV. It shows better performance Comparison of Multi hop Wireless Adhoc
performance after 1.5 m/s speed is because to two factors Network Routing Protocols”, Mobicomm’98, Texas, Oct
(i) Like AODV it triggers receives more replies and 1998.
11. E. Crawley, R. Nair, B. Rajagopalan, and H. Sandick, “A
requests and (ii) the routes becomes stable with better framework for QoS based routing in the internet,” RFC 2386,
selection of nodes. It has been observed that the end-to- Aug. 1998.
end throughput for all protocols decrease for high network 12. M. Ettus. System Capacity, Latency, and Power Consumption
speeds above 10 meter per second. in Multihop-routed SS-CDMA Wireless Networks. In Radio
and Wireless Conference (RAWCON ’98), pages 55–58, Aug.
5. Conclusion 1998.
13. X. Lin and I. Stojmenovic. Power-Aware Routing in Ad Hoc
Wireless Networks. In SITE, University of Ottawa, TR-98- 11,
A new scheme has been presented that utilizes a
Dec. 1998.
mesh structure and alternate paths. The scheme can be 14. A. Chockalingam and M. Zorzi, “Energy Consumption
incorporated into any ad hoc on-demand unicast routing Performance of a Class of Access Protocols for Mobile Data
protocol to improve reliable packet delivery in the face of Networks,” Proc. IEEE VTC, May 1998, pp. 820–24.
node movements and route breaks. Alternate routes are 15. A. Michail and A. Ephremides, “Energy Efficient Routing for
utilized only when data packets cannot be delivered Connection Oriented Traffic in Ad-hoc Wireless Networks,”
through the primary route. As a case study, the proposed Proc. IEEE PIMRC, Sept. 2000, pp. 762–66.
scheme has been applied to AODV and it was observed 16. G. Zussman and A. Segall. Energy efficient routing in ad hoc
that the performance improved. Simulation results disaster recovery networks. Proceedings of IEEE INFOCOM,
April, 2003.
indicated that the technique provides robustness to
17. C. Schurgers and M. B. Srivastava. Energy efficient routing
mobility and enhances protocol performance. It was found in wireless sensor networks. Proceedings of IEEE MILCOM,
that overhead in this protocol was slightly higher than pages 28–31, October 2001.
others, which is due to the reason that it requires more 18. C. K. Toh, “Maximum battery life routing to support
calculation initially for checking backbone nodes. This ubiquitous mobile computing in wireless ad hoc networks”,
also caused a bit more end to end delay. The process of IEEE Comm. Mag., June 2001, pp. 138-147.
checking the protocol scheme is on for more sparse 19. S. Singh, M. Woo and C. S. Raghavendra, “Power aware
mediums and real life scenarios and also for other metrics routing in mobile ad hoc networks”, IEEE/ACM MobiCom,
like Path optimality, Link layer overhead. Oct. 1998, pp. 181-190.
20. D. B. Johnson and D. A. Maltz, “Dynamic source routing in
ad hoc wireless networks”, Mobile Computing, Kluwer, 1996,
6. References pp. 153-181.

1. National Science foundation, “Research priorities in Wireless

and mobile networking”, Available at www.cise.nsf.gov .
2. E.M. Royer and C.K. Toh, “A review of current routing
protocols for ad hoc mobile wireless networks”. IEEE
Personal Communications, pages 46–55, April 1999.
3. Tsu-Wei Chen, Mario Gerla, "Global State Routing: A New
Routing Scheme for Ad-hoc Wireless Networks" Proceedings
IEEE ICC 1998.
4. S. Murthy, J.J. Garcia-Luna-Aceves, "An Efficient Routing
Protocol for Wireless Networks", ACM Mobile Networks and
App. Journal, Special Issue on Routing in Mobile
Communication Networks, Oct. 1996, pp. 183-97.
5. Z.J. Hass, M.R. Pearlman, “Zone routing protocol (ZRP)”,
Internet draft. June 1999, at www.ietf.org .
6. J.J. Garcia, M. Spohn and D. Bayer, “Source Tree Adaptive
Routing protocol”, IETF draft, October 1999.
7. D.B. Johnson, D.A. Maltz, "Dynamic Source Routing in Ad
Hoc Networks", Mobile Computing, T. Imielinski and H.
Korth, Eds., Kulwer, 1996, pp. 152-81.
8. G. Aggelou, R. Tafazoli, “Bandwidth efficient routing
protocol for mobile ad hoc networks (RDMAR)”, CCSR ,UK.
9. C.E. Perkins, E.M. Royer, “Ad-Hoc On Demand Distance
Vector Routing”, Proceedings of the 2nd IEEE Workshop on
Mobile Computing Systems and Applications (WMCSA),
New Orleans, LA, February 1999, pp. 90-100.

1033
1034