Content uploaded by Kalyan Kaipa
Author content
All content in this area was uploaded by Kalyan Kaipa on Jun 18, 2016
Content may be subject to copyright.
Energy Efficient Routing in Wireless Ad Hoc Networks
S.-M. Senouci and G. Pujolle
Laboratoire LIP6 – Université de Paris VI
8, rue du Capitaine Scott – 75015 Paris, France
{senouci, gp}@rp.lip6.fr
$EVWUDFW²$G KRF ZLUHOHVV QHWZRUNV DUH SRZHU FRQVWUDLQHG VLQFH
QRGHV RSHUDWH ZLWK OLPLWHG EDWWHU\ HQHUJ\ 7KXV HQHUJ\
FRQVXPSWLRQ LV FUXFLDO LQ WKH GHVLJQ RI QHZ DG KRF URXWLQJ
SURWRFROV7RGHVLJQVXFKSURWRFROVZHKDYH WRORRN DZD\IURPWKH
WUDGLWLRQDO PLQLPXP KRS URXWLQJ VFKHPHV ,Q WKLV SDSHU ZH
SURSRVHWKUHHH[WHQVLRQVWRWKHVWDWHRIWKHDUWVKRUWHVWFRVWURXWLQJ
DOJRULWKP $2'9 7KH GLVFRYHU\ PHFKDQLVP LQ WKHVH H[WHQVLRQV
/($5$2'9 3$5$2'9 DQG /35$2'9 XVHV HQHUJ\
FRQVXPSWLRQ DV D URXWLQJ PHWULF 7KH\ UHGXFH WKH HQHUJ\
FRQVXPSWLRQ RI WKH QRGHV E\ URXWLQJ SDFNHWV WR WKHLU GHVWLQDWLRQ
XVLQJ HQHUJ\RSWLPDO URXWHV :H VKRZ WKDW WKHVH DOJRULWKPV
LPSURYH WKH QHWZRUN VXUYLYDELOLW\ E\ PDLQWDLQLQJ WKH QHWZRUN
FRQQHFWLYLW\ 7KH\ FDUU\ RXW WKLV R EMHFWLYH ZLWK ORZ RYHUKHDG DQG
ZLWKRXWDIIHFWLQJWKHRWKHUZLUHOHVVQHWZRUNSURWRFROOD\HUV
.H\ZRUGVDG KRF URXWLQJ HQHUJ\ FRQVXPSWLRQ QHWZRUN
VXUYLYDELOLW\$2'9
I. INTRODUCTION
The increasing progress of wireless local area networks (WLAN) has
opened new horizons in the field of telecommunications. Among the
various network architectures, the design of mobile ad hoc network
(MANET) has attracted a lot of attention. A MANET is composed of a
set of mobile hosts that can communicate with one another. No base
stations are supported in such an environment, and mobile hosts
communicate in a multi-hop fashion. Such networks are needed in
situations where temporary network connectivity is required, such as in
battlefields, disaster areas, and meetings, because of their capability of
handling node failures and fast topology changes. Those networks
provide mobile users with ubiquitous communication capability and
information access regardless of location.
A set of ad hoc routing protocols have been proposed in the IETF’s
MANET [1] group to ensure the network connectivity. They operate in
either proactive or reactive modes. Proactive protocols are table-driven
and maintain routes for the entire network. Nodes must be in continuous
communication for updating changes in the topology. In reactive
protocols, a route to a destination is established only on demand, based
on an initial discovery between the source and the destination.
Building such routing algorithms poses a significant technical
challenge, since the devices are battery operated. The devices need to be
energy conserving so that battery life is maximized. The shortest path is
the most common criteria adopted by the conventional routing protocols
proposed in the MANET Working Group. The problem is that nodes
along shortest paths may be used more often and exhaust their batteries
faster. The consequence is that the network may become disconnected
leaving disparity in the energy, and eventually disconnected sub-
networks. Therefore, the shortest path is not the most suitable metric to
be adopted by a routing decision. Other metrics that take the power
constraint into consideration for choosing the appropriate route are more
useful in some scenarios (e.g. sensor networks).
In this paper, we propose three energy efficient routing algorithms
(LEAR-AODV, PAR-AODV, and LPR-AODV) that reduce energy
consumption and lead to a longer battery life at the terminals. They are
based on one of the most important routing protocols, AODV (Ad hoc
On-Demand Distance Vector) [2]. We focus on reactive routing
schemes, since they are less expensive in terms of energy consumption
than proactive schemes [3].
The remainder of this paper is organized as follows. We detail the
different energy efficient routing algorithms we propose in section 2.
Performance evaluation and numerical results are exposed in section 3.
Finally, section 4 summarizes the main contributions of this work.
II. ENERGY EFFICIENT ROUTING ALGORITHMS
In this section, we present new energy efficient routing algorithms.
They are designed to increase the network survivability by maintaining
the network connectivity and to lead to a longer battery life of the
terminals. This is in contrast to AODV which does not consider power
but optimizes routing for lowest delay. The protocols we developed
(LEAR-AODV, PAR-AODV, and LPR-AODV) ensure the
survivability of the network by establishing routes that ensure that all
nodes equally deplete their battery power. They are reactive protocols
and are based on the AODV routing protocol described below.
A. AODV Protocol
AODV routing protocol [2] is a reactive routing algorithm. It
maintains the established routes as long as they are needed by the
sources. AODV uses sequence numbers to ensure the freshness of
routes. Route discovery and route maintenance for AODV are described
below.
A.1. Route Discovery
The route discovery process is initiated whenever a traffic source
needs a route to a destination. Route discovery typically involves a
network-wide flood of route request (RREQ) packets targeting the
destination and waiting for a route reply (RREP). An intermediate node
receiving a RREQ packet first sets up a reverse path to the source using
the previous hop of the RREQ as the next hop on the reverse path. If a
valid route to the destination is available, then the intermediate node
generates a RREP, else the RREQ is re-broadcast. Duplicate copies of
the RREQ packet received at any node are discarded. When the
destination receives a RREQ, it also generates a RREP. The RREP is
routed back to the source via the reverse path. As the RREP proceeds
towards the source, a forward path to the destination is established.
A.2. Route Maintenance
Route maintenance is done using route error (RERR) packets. When
a link failure is detected, a RERR is sent back via separately maintained
predecessor links to all sources using that failed link. Routes are erased
by the RERR along its way. When a traffic source receives a RERR, it
initiates a new route discovery if the route is still needed. Unused routes
in the routing table are expired using a timer-based technique.
B. Local Energy-Aware Routing based on AODV (LEAR-
AODV)
The first on-demand routing protocol we propose is called LEAR-
AODV (Local Energy-Aware Routing based on AODV). The main
objective is to balance energy consumption among all participating
nodes. We use a similar mechanism to that used in [2], where the
authors propose to extend the DSR (Dynamic Source Routing) [4]
protocol. In their approach, each mobile node relies on local information
about the remaining battery level to decide whether to participate in the
selection process of a routing path or not. An energy-hungry node can
conserve its battery power by not forwarding data packets on behalf of
others. The decision-making process in LEAR-AODV is distributed to
all relevant nodes. Route discovery and route maintenance for LEAR-
AODV are described below.
B.1. Route Discovery
In AODV, each mobile node has no choice and must forward packets
for other nodes. In LEAR-OADV, each node determines whether or not
to accept and forward the RREQ message depending on its remaining
battery power ((
). When it is lower than a threshold value
θ
((
θ
), the
RREQ is dropped; otherwise, the message is forwarded. The destination
will receive a route request message only when all intermediate nodes
along the route have enough battery levels.
B.2. Route Maintenance
Route Maintenance is needed either when the connections between
some nodes on the path are lost due to node mobility, or when the
energy resources of some nodes on the path are depleting too quickly. In
the first case, and as in AODV, a new RREQ is sent out and the entry in
the route table corresponding to the node that has moved out of range is
purged. In the second case, the node sends a route error RERR back to
the source even when the condition (
θ
is satisfied. This route error
message forces the source to initiate route discovery again. This is a
local decision since it is dependent only on the remaining battery
capacity of the current node.
However, if this decision is made for every possible route, the source
will not receive a RREP message even if there exists a route between
the source and the destination. To avoid this situation, the source will re-
send another RREQ message with an increased sequence number.
When an intermediate node receives this new request, it lowers it’s
θ
by
d to allow the packet forwarding to continue. We use a new control
message, $'-867B7KU. When a node drops a RREQ message, it
instead broadcasts a $'-867B7KU message. The subsequent nodes
closer to the destination now know that a request message was dropped
and lower their threshold values. Now, the second route request
message can now reach the destination. When the destination receives a
RREQ, it generates a RREP. As in AODV, the RREP is routed back to
the source via the reverse path.
We notice that LEAR-AODV interworks easily with AODV. By
this, we mean that an ad hoc network can contain both nodes carrying
out LEAR-AODV, and nodes carrying out AODV as routing protocol.
C. Power-Aware Routing based on AODV (PAR-AODV)
The second on-demand routing protocol we propose is called PAR-
AODV (Power-Aware Routing based on AODV). The main objective
is to extend the useful service life of an ad hoc network. PAR-AODV
solves the problem of finding a route
π
, at route discovery time W, such
that the following cost function [6] is minimized:
∑
∈
=
π
π
W&W& )(),( , (1)
where
α
ρ
=)(
)( W(
)
W&
(2)
and
ρ
is the transmit power of node L;
)
is the full-charge battery capacity of node L;
(
W is the remaining battery capacity of node L at time W;
α
is a positive weighting factor.
The route discovery and route maintenance for PAR-AODV are
described below.
C.1. Route Discovery
In PAR-AODV, activity begins with the source node flooding the
network with RREQ packets when it has data to send. All nodes except
the source and the destination calculate their link cost, &
, using (2), and
add it to the path cost in the header of the RREQ packet (cf. (1)). When
the destination node receives a RREQ packet, it sends a RREP packet to
the source.
When an intermediate node receives a RREQ packet, it keeps the cost
in the header of that packet as 0LQ&RVW. If additional RREQs arrive
with the same destination and sequence number, the cost of the newly
arrived RREQ packet is compared to the 0LQ&RVW:
- If the new packet has a lower cost and if the intermediate node does
not know any valid route to the destination, 0LQ&RVWis changed to this
new value and the new RREQ packet is re-broadcast;
- If the new packet has a lower cost but the intermediate node knows
a route to the destination, the node forwards (unicast) a
&20387(B&RVWmessage. The &20387(B&RVWcalculates this route
cost;
- Otherwise, if the new packet has a greater cost, the new RREQ
packet is dropped.
When the destination receives either a RREQ or a &20387(B&RVW
message, it generates a RREP message. The RREP is routed back to the
source via the reverse path. This reply message contains the cost of the
selected path. The source node will select the route with the minimum
cost.
C.2. Route Maintenance
The route maintenance in PAR-AODV is the same as in LEAR-
AODV. Hence, in PAR-AODV, when any intermediate node has a
lower battery level than its threshold value ((
θ
), any request is simply
dropped.
D. Lifetime Prediction Routing based on AODV (LPR-AODV)
The last on-demand routing protocol we propose is called LPR-
AODV (Lifetime Prediction Routing based on AODV). This protocol
favors the route with maximum lifetime, i.e. the route that does not
contain nodes with a weak predicted lifetime. LPR-AODV solves the
problem of finding a route π at route discovery time W, such that the
following cost function is maximized:
(
)
(
)
∈
=)()( W
L
7
L
0LQ0D[W70D[
π
π
π
π
(3)
where 7 W is the lifetime of path
π
;
7
W is the predicted lifetime of node L in path
π
.
LPR-AODV uses battery lifetime prediction. Each node tries to
estimate its battery lifetime based on its past activity. This is achieved
using a recent history of node activity. When node L sends a data packet,
it keeps track of the residual energy value ((
(W)) and the corresponding
time instance (W). This information is recorded and stored in the node.
After 1 packets sent/forwarded, node L gets the time instance when the
1th packet is sent/forwarded (W¶) and the corresponding residual energy
value ((
(W¶)). This recent history, {(W, (
(W)), (W¶, (
(W¶))}, is a good
indicator of the traffic crossing the node. Hence, we use it for lifetime
prediction. Our approach is a dynamic distributed load balancing
approach that avoids power-congested nodes and chooses paths that are
lightly loaded. Route discovery and route maintenance for LPR-AODV
are described below.
D.1. Route Discovery
In LPR-AODV, all nodes except the destination and the source
calculate their predicted lifetime, 7
, using (4). In each request, there is
another field representing the minimum lifetime (0LQOLIHWLPH) of the
route. A node L in the route replaces the 0LQOLIHWLPHin the header with
7
if 7
is lower than the existing 0LQOLIHWLPHvalue in the header.
)(ratedischarge_
)(
(t)
i
TW
L
W
L
(
= (4)
where
’
)()’(
)(
i
ratedischarge_
W
W
W
L
(W
L
(
W
−
−
=
and (
(W) is the remaining energy of node L at time W;
W: current time corresponding to the moment when the node L
sends/forwards the current packet;
W: the recorded time instance corresponding to the moment when the
1th ‘predecessor’ to current packet was sent/forwarded by node L.
More precisely, when an intermediate node receives the first RREQ
packet, it keeps the 0LQOLIHWLPH in the header of that packet as 0LQ
/LIHWLPH. If additional RREQs arrive with the same destination and
sequence number, the 0LQOLIHWLPHof the newly arrived RREQ packet is
compared to the 0LQOLIHWLPH:
- If the new packet has a greater 0LQOLIHWLPHand if the intermediate
node does not know any valid route to the destination, 0LQOLIHWLPHis
changed to this new value and the new RREQ packet is re-broadcast;
- If the new packet has a greater 0LQOLIHWLPHbut the intermediate
node knows a route to the destination, the node forwards (unicast) a
&20387(BOLIHWLPHmessage. The &20387(BOLIHWLPHcalculates this
route lifetime;
- Otherwise, if the new packet has a lower 0LQOLIHWLPH, the new
RREQ packet is dropped.
When the destination receives either a RREQ or a
&20387(BOLIHWLPH message, it generates a RREP message. The
RREP is routed back to the source via the reverse path. This reply
message contains the lifetime of the selected path. The source node will
select the route with the maximum lifetime.
D.2. Route Maintenance
As in the first algorithms, route maintenance is needed either when a
node becomes out of direct range of a sending node or there is a change
in its predicted lifetime. In the first case (node mobility), the mechanism
is the same as in AODV. In the second case, the node sends a route error
RERR back to the source even when the predicted lifetime goes below a
threshold level
δ
(7
(W)
δ
). This route error message forces the source to
initiate route discovery again. This decision depends only on the
remaining battery capacity of the current node and its discharge rate.
Hence, it is a local decision.
However, the same problem as in LEAR-AODV can occur. If the
condition 7
(W)
δ
is satisfied for all the nodes, the source will not receive
a single reply message even though there exists a path between the
source and the destination. To prevent this, we use the same
mechanisms used in LEAR-AODV described above.
III. EXPERIMENTAL RESULTS
The performances of our algorithms are evaluated using GloMoSim
2.0 simulator [8]. The simulation consists of a network of 36 nodes
confined in a 800×800 m² area. Random connections were established
using CBR traffic (at 4 packets/second with a packet size of 1024
bytes). The initial battery capacity of each node is 10 units. This initial
energy is progressively reduced by data transmission/reception. When it
reaches zero units, the corresponding node cannot take part any more in
the communication, and is regarded as died. Each node has a radio
propagation range of 250 meters and channel capacity was 2 Mb/s. We
consider the simple case when the transmit power is fixed. In this case,
each packet relayed or transmitted consumes a fixed amount of energy
from the battery.
The performance metric, in these kind of studies, is the network
lifetime. The network lifetime can be defined as [6]:
- the time taken for K nodes in the network to die;
- the time taken for the first node to die;
- the time for all nodes in the network to die.
In this work, we adopt the first and second definitions. Network
lifetimes of our algorithms are compared for different scenarios. They
are often compared to AODV since they are derived from it. Two cases
were considered: (i) the nodes are fixed, and (ii) the nodes are mobile
and move with various velocities.
A. Fixed Nodes
Figure 1 shows the time instances at which certain number of nodes
has died because of their batteries depletion, when all the nodes are
fixed. We note that for AODV, the first node dies approximately 2056
seconds earlier than in LEAR-AODV, 2572 seconds earlier than in
PAR-AODV, and 3244 seconds earlier than in LPR-AODV. Similarly,
and for 4 nodes, those die approximately 888 seconds earlier than in
LEAR-AODV, 1132 seconds earlier than in PAR-AODV, and 1832
earlier than in LPR-AODV.
0
2000
4000
6000
8000
1 2 3 4 5 6 7
LEAR-AODV
PAR-AODV
AODV
LPR-AODV
Figure 1. The number of dead nodes versus time.
LPR-AODV is better than PAR-AODV since LPR-AODV takes
into account not only the residual battery capacity, but also the rate of
energy discharge. We have carefully compared the performance of
LPR-AODV for different values of N, and found 1=5 to be a good
value.
AODV
0100 200 300 400 500 600 700 800
simulation area 0100200 300 400 500 600 700 800
0
2
4
6
8
10
(a)
LEAR-AODV
0100 200 300 400 500 600 700 800
simulation area 0100200 300 400 500 600 700 800
0
2
4
6
8
10
(b)
Figure 2. Battery levels of all the ad hoc network nodes using (a) AODV (b)
LEAR-AODV.
In order to improve the survivability of the network, the variance of
energies between all the nodes should be reduced to the minimum.
Figure 2 gives the battery levels of all ad hoc nodes after a total
simulation time of 1000 seconds. We consider for this experiment that
the nodes are fixed. We represent the results of LEAR-AODV (cf.
Figure 2(b)), compared to those of AODV (cf. Figure 2(a)). In LEAR-
AODV, the nodes consume energy more equitably. Thus, the nodes in
the center of the network continue to maintain the network connectivity
as long as possible, and the network will not be partitioned rapidly. On
the other hand, for AODV, the energy level of the nodes in the center is
largely lower than the half of the initial energy level.
B. Mobile Nodes
The effect of mobility is shown in Figure 3. As can be seen our
algorithms are always better than AODV in terms of dead nodes. We
note that for AODV, and for a node velocity equal to 4 meters/second
for example, the first node dies approximately 793 seconds earlier than
in LEAR-AODV, 1125 seconds earlier than in PAR-AODV, and 1182
seconds earlier than in LPR-AODV. However, as the velocity of the
node movement increases, rate of energy consumption in the network
goes up. This is normal since higher velocity of movement implies more
route discoveries being performed and as a consequence higher energy
consumption in the network. Also, as the node mobility increases, the
difference between AODV and our algorithms decreases. Because there
are more route discoveries, no paths are overused even by AODV. As a
consequence, AODV also achieves load balancing to an extent
decreasing the gain seen by our algorithms.
0
1000
2000
3000
4000
5000
6000
012345678
Time (sec)
Number of dead nodes (velocity is 1m/s)
AODV
PAR-AODV
LEAR-AODV
LPR-AODV
(a)
0
1000
2000
3000
4000
5000
6000
012345678
Time (sec)
Number of dead nodes (velocity is 4m/s)
AODV
PAR-AODV
LEAR-AODV
LPR-AODV
(b)
0
1000
2000
3000
4000
5000
6000
012345678
Time (sec)
Number of dead nodes (velocity is 8m/s)
AODV
PAR-AODV
LEAR-AODV
LPR-AODV
(c)
Figure 3. Number of dead nodes with a velocity of (a) 1m/s (b) 4m/s (c) 8m/s.
In these algorithms, route discovery process needs more control
packets to be propagated in the network. To show the overhead of our
algorithms, we have measured the ratio of the size of all the control
packets to the size of all data packets delivered in the network. Figure 4
shows this ratio for our algorithms for different velocities of node
movement, with a simulation time of 6000 seconds. As the velocity of
movement increases, routes are valid for shorter time and more route
discoveries are done in the network resulting in more control packets
and more the difference between the algorithms.
0
2
4
6
8
10
0 1 4 8
Overhead (%)
Velocity of node movement (m/s)
AODV
LEAR-AODV
PAR-AODV
LPR-AODV
Figure 4. Bytes overhead as function of velocity of node movement.
IV. CONCLUSION
One critical issue for almost all kinds of portable devices supported
by batteries is power saving. Without power, any mobile device will
become useless. Battery power is a limited resource, and it is expected
that battery technology is not likely to progress as fast as computing and
communication technologies do. Hence, how to lengthen the lifetime of
batteries is an important issue, especially for MANET, which is all
supported by batteries.
Routing and power consumption are intrinsically connected. In
conventional routing algorithms, which are unaware of energy budget,
connections between two nodes are established through the shortest
routes. These algorithms may however result in a quick depletion of the
battery energy of the nodes along the most heavily used routes in the
network.
In this paper, we design new power-aware routing protocols (LEAR-
AODV, PAR-AODV, and LPR-AODV) that balance the traffic load
inside the network so as to increase the battery lifetime of the nodes and
hence the overall useful life of the ad hoc network. These protocols are
based on the conventional AODV.
These AODV extensions increase the network survivability and lead
to a longer battery life of the terminals. They achieve balanced energy
consumption with minimum overhead. Simulation results show that our
algorithms increase clearly network lifetime. Another important
advantage of these algorithm is their simplicity and the fact that they do
not affect other layers of wireless communication protocols.
REFERENCES
[1] IETF MANET WG (Mobile Ad hoc NETwork),
www.ietf.ora/html.charters/manet-charter.html.
[2] C. E. Perkins, E. M. Belding-Royer, and S. R. Das, “Ad hoc On-Demand
Distance Vector (AODV) Routing”,
!#" $%&!'$("*)+%&, -("
, draft-ietf-manet-
aodv-13.txt.
[3] L. Ouakil, S. Senouci, and G. Pujolle, "Performance Comparison of Ad Hoc
Routing Protocols Based on Energy Consumption",
./10'2
$!'34$6517(%&89&:;7 <
=?>'>=
, Torino, Italy, September 2002.
[4] D. B. Johnson, D. A. Maltz, Y.–C. Hu, “The Dynamic Source Routing Protocol
for Mobile Ad Hoc Networks (DSR)”,
@ !#" $%&!'$("#)+%&, -"
, draft-ietf-manet-
dsr-09.txt.
[5] K. Woo, C. Yu, D. Lee, H. Y. Youn, and Ben Lee, “Non-Blocking, Localized
Routing Algorithm for Balanced Energy Consumption in Mobile Ad Hoc
Networks”,
AB.C#DE
CGF
>H
, Cincinnati, Ohio, August 15-18, pp. 117-124,
2001.
[6] M. Maleki, K. Dantu, and M. Pedram, “Power-aware Source Routing in mobile
ad hoc networks”,
I
%J734$$K
2
!(L;9M7 -N
CPOQI
G)SR >P=
, Monterey, CA, pp. 72-75,
August 12-14, 2002.
[7] M. Maleki, K. Dantu, and M. Pedram, “Lifetime Prediction Routing in Mobile
Ad Hoc Networks”,
GGT5
2
%&$(U $9&9
D
7
/V/+W
!
2
34,;"
2
7P!X,'!KZY*$(" [*7(%&8
2
!(L
D
7P! -P\
,
Mars, 2003.
[8] X. Zeng, R. Bagrodia, and M. Gerla, “GloMoSim: a Library for Parallel
Simulation of Large-scale Wireless Networks”,
IQ.
)
C]F^'_
, Banff, Alberta,
Canada, May 26-29, 1998.
