2012 2nd IEEE International Conference on Parallel, Distributed and Grid Computing

Energy Conscious DSR in MANET Baisakh Nileshkumar R Patel Shishir Kumar

Sambalpur University of Information Technology,

Jaypee University of Engineering and Technology,

Jaypee University of Engineering and Technology,

Sambalpur, India baisakh@suiit.ac.in

Guna, India nilesh.kumar@juet.ac.in

Guna, India shishir.kumar@juet.ac.in

Abstract- now days, the versatile use of Mobile Ad Hoc

Networks (MANET) have elicited everyone's attention and offer

a lot of challenges to the researchers. Due to many salient

features of MANET and the absence of base stations or central

coordinators, all the actions are been played by the individual

node. One of the major responsibility is routing which has to be

taken up by the node only, hence the routing mechanism is been

incorporated in the node itself. Usually routing protocols play a

very serious role in navigating the packets through many

intermediate nodes from source to destination. Hence it is a very

costly process which consumes a lot of power. But many

traditional routing protocols don't concern about the energy

consumption of the individual node and they select the path on

the basis of minimum hop count. Very often the minimum hop

count leads to high power consumption. And as the traditional

routing protocol don't bother about the node residual energy, in

any time if a single intermediate node dies out of energy then the

whole communication may be interrupted due to network

partition which may also increase the number of retransmission

of packet, packet loss, consume high bandwidth and hence gives

poor performance. Because all these nodes are operated by the

limited battery power it has to be used efficiently so that the life

time of the network can be enhanced without degrading the

performance. In this paper we have proposed an Energy

Conscious routing protocol by modifying one of the most popular

routing protocols that is Dynamic Source Routing (DSR) protocol

which is not at all concerned about power consumption. This

Energy conscious DSR (ECDSR) uses the basic concept of

traditional DSR and imposes it's two importance characteristics

as Energy saving and Energy Survival in DSR, through which it

not only enhances the life time of the network but also increases

the overall performance of the networks. We have validated our

proposed protocol through ns-2.34 and evaluated the

performance of the networks considering few energy metrics. We

found that our proposed method ECDSR outperforms DSR in

the performance analysis.

Index Terms-MANET, energy consumption, DSR, ECDSR

MOBILE AD HOC NETWORKS (MANET)

A Mobile Ad Hoc Network (MANET) is an autonomous system of mobile nodes where the nodes where the nods are equipped with wireless transmitters and receivers using antennas which may be unidirectional (broadcast), highly­directional (point-to-point) and a special routing protocol. So depending upon the node's location, inter node distance and their receiving and transmitting power levels, a wireless connection is established among the participated nodes to form an "Ad-Hoc" network [ I, 2] . This Ad-hoc topology may

978-1-4673-2925-5/12/$31.00 ©2012 IEEE 784

change with time as the nodes move or adjust their transmission and reception parameters. As MANET is also a wireless network, it possesses some common issues as the wireless network is having, like bandwidth optimization, power control and transmission quality enhancement. But due to MNAET several salient features, it has also introduced many new challenges like auto configuration, router mechanism, security, scalability and many more. And the most important constraint is that while making out solutions to those above new challenges we cannot ignore power which is being used during the whole process. So power consumption in MANET has become the ultimate problem which is implicitly or explicitly associated with all other problems as the nodes are operated by the limited battery power. In such scenario, selecting a suitable routing protocol which would not lead to any serious issues mention above is important while selecting a path from a source to destination.

Energy consumption in MANET

As no centralized infrastructure exists in MANET, the routing mechanism has been incorporated in individual nodes. And due to the dynamic nature of MANET, it makes even more difficult for the individual node to establish most suitable path from source to destination and for each source to destination, it has to keep track of the established route in regular time of interval. If any one of the internal nodes dies out of energy while the communication is established then the whole process will be interrupted and again the path has to be chosen for the specified source to destination. So this regular route updation and route maintenance consumes a lot of energy of the battery which is limited. More ever, the traditional routing protocol of MANET is one of the major causes of high power consumption. Because most of the traditional routing mechanism is based on minimum hop count i.e. while selecting a path from as source to any destination, it selects the only path which is having minimum number of intermediate nodes among all possible paths. As the distances between each pair of nodes get increased, the amount of transmission power also gets increased. And it is found that the more amount of energy is consumed while transmitting a data rather than receiving a data. So the power level of nodes are also getting affected the ease with which route is established between a pair of nodes. Apart from this issue, many traditional routing protocols are least concerned with the energy consumption of the nodes and few energy related parameters like energy consumed per packet, energy required per transmission, remaining battery power of the node etc.

2012 2nd IEEE International Conference on Parallel, Distributed and Grid Computing

So in this paper, we have consider the basic mechanism of one of the well known routing protocol that is Dynamic Source Routing (DSR) protocol and we have made an attempt to modify this DSR as an Energy conscious DSR (ECDSR). We have chosen DSR because it has been accepted as a better routing protocol with compare to any other protocols like AODV, TORA by considering various performance metrics.

So the next section II will shade some lights on basic DSR mechanism and it reveals why DSR requires some more attention in order to keep its position as a better routing protocol. Then in the session III we have explained the proposed method that is ECDSR and its mechanism. Then we have analyzed some results of the performed simulation and analyzed briefly in the session IV.

DYNAMIC SOURCE ROUTING

Basically in MANET the routing protocols [ 13, 14] have been divided into two categories, one is proactive and another is reactive protocol. In the proactive protocols every node maintains the route for the entire networks where as in the reactive protocol; a route to a certain destination is established only when there is a demand from a source for a particular destination. DSR comes under reactive protocol which is a well known protocol among all reactive protocols. DSR primarily performs two basic operations during its whole routing process, which are route discovery and route maintenance. If a node wants to send a packet to a destination, then prior to that it has to check its own routing cache whether there is any existing route available to that destination or not. If it finds a route, then it uses it to send the packet to the destination. Otherwise it starts the route discovery process.

Route Discovery

In the route discovery phase, a source starts looking for a route from source to destination when it has some data packet to send. First the source node checks its own route cache as every node is having its own route cache and if it does not find the route to the destination then it goes for another two sub phases, one is Route Request and Route Reply.

Route Request When the route is not found in the route cache of the source node, then the source node broadcasts a Route Request (RR EQ) message to all its neighbor nodes by attaching its own address, destination address and a unique identification number so that each node processes the RR EQ only once. Each intermediate node appends its own address to the route record of the packet and keeps broadcasting the RREQ message until it finds the destination node in its own route cache during the RR EQ propagation.

Route Reply A route reply (RR EP) message is generated by the destination node when the destination node receives the RR EQ with same destination address in the RREP packet or being generated by any an intermediate node if the routing information about the destination is available in its own route cache. Mean while the sequence of hops are kept updating in the RR EQ packet. On

785

the basis of minimum hop count, the destination node selects the best path and generates the RR EP packet back to the source by placing the route record from route request packet into the route reply packet. In case of an intermediate node, when it generates the RR EP message then it appends its cached route to the destination to the route record in the RR EQ packet and then generates the RR EP packet. Finally, the source accumulates the route carried by the RR EP that it receives for future use.

A. Route Maintenance

MANET should have certain mechanism to maintain the linl<­from source to destination due its dynamic in nature. At any particular time if a single participating node comes out of the range of its neighbor node then it may lead to a network partition. A network partition not only increases the packet loss but also compel to begin few costly operations like route discovery, route maintenance, path repair and many more. The route maintenance is performed by generating route error message or through different acknowledgement. So in general, route maintenance acts as a route repair phase once some trouble in linl<- occurs while sending the packets.

B. Benefits and Limitation

As the packet themselves contains the routing decision, the intermediate nodes do not need to maintain the up-to-date routing information. More ever, with the caching mechanism in of any initiated or overheard routing data also reduces the number of control message being sent which in turn reduces the overhead.

But still it has few major limitations. As DSR routing mechanism is based on minimum hop counts, the distance between each pair of nodes becomes larger and hence higher transmission power is required for the communication. As the transmission power becomes high, more energy will be consumed. DSR is also considered as one of the unconventional routing protocols as it is least concerned about the energy related parameters which are more crucial for any mobile ad hoc networks. If any advancement and new technology are used to improve the performance of the networks at the same time it should not increase the rate of energy consumption at node level. The more the depletion rate of the battery, the less will be the life time of the node, the less will be the life time of the networks and hence more will be the network partition. If the degree of network partition will be higher then the number of retransmission will be higher and throughput will be gradually reduced.

Hence while proposing the new method, we have emphasized more on the battery power consumption and the overall performance of the system which should not be degraded as the DSR gets changed to energy conscious DSR.

II. ENERGY CONSCIOUS DSR (ECDSR)

The primary objective of ECDSR is to select the path for the specified source to destination in such a way that all intermediate nodes will have higher level of energy at a given time. So instead of following minimum hop count method during the route discovery phase, select those path whose

2012 2nd IEEE International Conference on Parallel, Distributed and Grid Computing

intermediate nodes are having higher remaining battery power, The moment, energy level of any node reaches to the minimum threshold value, a new path will be chosen in order to continue the communication without any interruption. So the modified algorithm has not only introduced an energy saving feature but also introduced an energy survival characteristic for any low energy node. Hence our proposed algorithm is more likely to be energy conscious while establishing a connection through many intermediate nodes. So ECDSR has two main phases, one is ECDSR Energy Saving Phase and the other one is ECDSR Energy Survival Phase.

A. ECDSR Energy Saving Phase

The energy saving mechanism is applied during the route discovery phase when the source node has some packets to transmit to the destination and it wants to have the most energy efficient path among all possible path, so that the communication can be carried out for a long period of time. So the route discovery phase of traditional DSR has been changed from minimum hop count to maximum energy conscious path carrying higher energy intermediate nodes.

[n the traditional DSR during route discovery process if any node receives a Route Request (RR EQ) which is not to be meant for it or if the receiving node is not the fmal destination then it holds the packet for a certain time interval which has been selected pseudo-randomly from a uniform distribution of probability between 0 to 0.0 l. Usually 0.0 1 is considered as constant broadcast jitter, After keeping the packet for that amount of time, it then broadcasts the packet.

But in ECDSR, we have introduced a kind of dynamic jitter for the control packets like RREQ and RR EP. And the method of using different value of jitter depends on the remaining residual battery power of the intermediate nodes. The value of jitter has not been disturbed as the range of jitter varies from 0 to 0.0 [ sec. The more will be residual power of the intermediate nodes, the lesser amount of delay will be imposed on the control packets. If any node is having low energy than certain threshold value, then it will avoid itself from participating as an intermediate node and sends an error message to the source. The delay in our proposed method is inversely proportional to the residual battery power, As in the traditional DSR, the pseudo-random delay and maximum of 0.0 1 sec, without violating it, ECDSR has used a special formula to calculate the delay. If the energy of a node is more than the minimum level of energy that is lJoule (low energy), then the delay should be less than 0.0 1 sec. Otherwise the delay will be maximum which is O.O[ second . The minimum value of the energy is set to 1 joule and it corresponds to the delay of II ([ * [00) = 0.0 1 sec. [n our simulation we have set the initial energy of the node as 10juole. So the delay introduced at RR EQ packet is II ([0*[00) =0.00 1 sec. As the node is been used gradually, the initial energy value will vary that is from 10 J to lJ.

So the new path in ECDSR contains the path with maximum some of total remaining energy of the intermediate nodes which in turn provides a higher possibility of a longer communication without any interruption of link breakage

786

unless the any intermediate node moves out of the range of its neighbor nodes. So with the above approach, the RREQ packet will reach at the destination which has maximum sum of remaining energy, because delay is inversely proportional to the remaining residual battery power.

B. ECDSR Energy Survival Phase

The energy survival phase comes into action during the route maintenance process. In DSR, link lost or retransmission of packet which may be caused by low residual power is never been considered. [f any time there was packet lost or route lost, either congestion or node mobility was taken as one of its reason. As in our experiment we have assumed our network to be static one, so we have neglect the node mobility possibility which may be one of the reasons of path lost or packet retransmission. But in ECDSR energy survival method, an additional cause is included where low residual energy might be one of the reasons of route breakage. And through this method we could also able to maintain a balance in energy consumption in all intermediate nodes.

As all intermediate nodes get used continuously unless communication gets over, the residual battery power is getting depleted continuously. So the moment any intermediate node finds its own residual energy level below or equal to a specific threshold value, then certainly it will inform to all its neighbor nodes that it is no more ready either to receive or to forward the packet. This can be informed through an alert message generated by the same node and insists to the source node to find another node for the same destination. So we have set the minimum threshold value as 1 Joule, so once any intermediate node has its remaining energy as 1 joule, it means it will set a "Low Energy" field to be [ and this field is to be added with the error control message. The moment neighbor nodes receive this error message with Low Energy field with 1, and then they will simply discard the route containing the affected node. And once this error message reaches at source, then the source starts looking for a new path present in its route cache, otherwise it will go for a route discovery process. So in this way, a node informs the source node to find a new path before any route breakage occurs and hence the established communication continues without any interruption. One other major advantage of this energy survival is to save the node from getting dead completely. It keeps the low energy node alive which can also be used in some other crucial communication in future.

C. ECDSR Algorithm

While developing this algorithm we have assumed that the nodes that are participating in the communication do not need to have equal initial battery power, Because we cannot assure that the node participating for a communication is not being used earlier.

1) Processing o/route request by source node: To fmd the efficient route, the route request is broadcasted

to the medium. The nodes which receive the route request packet compute the delay and add it to the packet header, So the route request can be processed either by the destination node or intermediate nodes.

2012 2nd IEEE International Conference on Parallel, Distributed and Grid Computing

2) Processing of the route request by destination node,' The node checks whether RR EQ is first arrived by looking up the sequence number and source node id in cache, If RR EQ is first arrived, the destination node sends a "Route Reply" to the initiator of the route request packet in which it includes the entire source route from the initiator to the destination. And then it will discard the other route request having the same source node or if the route request is having the same sequence number. The destination node will treat the route request as the first come first serve basis.

3) Processing of the route request by destination node: It checks whether RR EQ is new by looking up the sequence

number and source node id in cache as like in target nodes. If the RR EQ is new one then it calculates the delay and includes it in the packet header and after keeping the packet till the calculated delay time it forwards to its neighbor node. If the destination node is in the cache then simply it uncast the RREQ for checking. If the destination node is not in its route cache then it again broadcast the RR EQ to its neighbor nodes.

4) Processing of error control packets: Whenever the energy of any node reaches less than or equal to 1 Joule then "Low Energy" field in the DSR header packet

will be set to 1 and it is broadcasted to it's all neighbor nodes. The DSR header packet is shown in Table- I. When the neighbor nodes of the affected node receive the error packet, then they will remove the path containing the affected node from its route cache and broadcast an error (Route Error) to the source. As soon as the error message received by the source node, it searches the path which is present in its route cache or else it triggers the route discovery phase.

T bilE a e rror pac k fI t f ECDSR et orma 0 IP DSR DSR Unreachable DSR DSR Low

Header Fixed Route Node Source Source Energy Header Error Address Route Route

Header Header Address

[L.N]

I I I. EXPERIMENTAL RESULTS

We have used the network simulator-2.34 to perform our experiment. We have included the Energy Model as well as the Two Ray Ground Model to calculate the residual battery power and to fix up the transmission power at every node. We set up a static network consists of a test bed of 1 1 nodes confined in a 300m x 200m in rectangular area. All nodes are assigned with the number from Node-O to Node- 1O.

As we have set different initial battery power at different nodes, all even nodes are having 20Joules and odd numbered nodes are having 10 joules of initial battery energy. The length of simulation time is 6Osec. We have introduced FTP traffic at node-O which is a source node. Below the list of parameter are shown in the simulation table 2.

787

Table 2· Simulation parameters value

Table Simulation Parameters Value Head

1 Network Size 300x200 in meter

2 Number 0 Nodes 1 1

3 Simulation Duration 60sec

4 Traffic type FTP

5 Packet Size 1080 (TCP), 40 (ACK) in bytes

6 Queue Type Drop Tail

7 Propagation Model Two Ray Ground

8 Antenna Unidirectional

9 Transmission Range 250m

10 rxPower 0.925W

1 1 txPower 1.43 W

12 sleepPower 0.045W

13 Initial Energy of nodes 10 Joules 1,3,5,7,9

14 Initial Energy of nodes 20Joules 0,2,4,6,8, 10

A. Evaluation metrics and simulation results

The metrics used to evaluate the performance of the ECDSR and DSR routing protocols are packet delivery fraction, energy consumption per delivery of packet, network life time, remaining residual battery power of the node, number of dropping nodes and throughput.

Packet Delivery Fraction

� 100 �----�-----�------, � Original OSR POF � c 90 ECOSR POF -

:j :� c

.::: 70 ... :;l 60 '­LI.. :n '­., :. ..... ..... ., co ... ., -'" " ., 0.. o L-____ �L-____ � _____ �

" ., ..... :::l Q ..., c

;,:; c Q ..... ... 0.. 0:: :::l " C Q " on 0.0 '-., c ....

OSR ECOSR

Protocol

Figure 1: Packet delivery fraction

Energy consunption for successful data delivery

0.35

0.3

0.25

0.2

0.15

0.1

0.05

0

Original DSR energy onsunption � ECDSR energy consunption .....

OSR ECOSR

Protocol

Figure 2: Energy consumption for successful data delivery

60 "" �

"Ij c: 50 0 (l a) � c: 40

• .-1 '"

a) I:

'.-1 30

... a)

... • .-1 20 ....I

.;:,t. L. 0 10 :J ... a) Z

0

6

� 5 a) "Ij 0 c: 0.0 4 c:

'.-1 � � 0 3 L.

"Ij ... 0 2 L. a)

..0 I: ::l z

1

0

2012 2nd IEEE International Conference on Parallel, Distributed and Grid Computing

Node HUMber vs ReMaining Node Energlj

Original DSR Renaining node energ'J I1:Si"iZTI EC[)SR ReMining node energ'J _

Node nUl'lber

Figure 3: Node residual battery energy

Network lifetine

Original DSR Network lifetine � ECDSR Network lifetine .....

DSR ECDSR

Protocol

Figure 4: Network life time

Nunber of dropping nodes

Original DSR Dropping nodes [ZZ2D ECDSR Dropping nodes .....

DSR ECDSR

Protocol

Figure 5: Number of dropping nodes

788

11

10

9

8 ... ::l 7 �

..c: 0.0 6 ::l 0 L. 5 ..c: ...

4

3

2

1

Throughput of the Network

Original DSR throughput � ECDSR throughput .....

DSR ECDSR

Protocol Figure 6: Throughput of the network

B. Summary of ECDSR and DSR Results

1) Packet deliveryfraction The objective of this metric is to analyze one of the network performance metrics called Packet delivery fraction (PDF) which helps to analyze the loss rate of the network. ECDSR provides better performance than the original DSR. PDF of ECDSR is 85.3%, where as for DSR it is 82.7% which is clearly shown in the Figure- I. Certainly, ECDSR proves considerable improvement in this performance metric in our experiment. The reason of having the better PDF is due to survival instinct of the node which informs the destination node to find an alternative route for the same source to the destination before the path gets broken. So before the new path is been selected, the data keeps moving through the same path. It reduces the number of retransmission and reduces data lost.

2) Energy consumption per successful delivery of data From the Figure-2 it is observed that the traditional DSR is consuming more energy per packet that is 0.29 Joules, whereas ECDSR consume 0.28 louie lesser than that of DSR. Even though it is small difference but for a network with a higher number of packet transmissions, it will have a high impact on the network life time as well as on individual node's life time. For example, in DSR every 1000 packets will consume 290 Joule of energy, whereas in ECDSR, 1000 packets will consume 280 Joule of energy. So for 1000 packets we are able to reduce 100 Joule energy which will really have a significant impact on overall network lifetime.

3) Remaining residual battery of the nodes

From the figure-3 it is shown that the node 1, 3, 5, and 7 are completely used and their energy level became zero. Because these are the only intermediate nodes through which data communication is possible. Note that an intermediate node can be replaced by some another node in the method 0 energy survival only when any alternative route is found, otherwise data will keep lowing through the same low energy intermediate node. The rest of the nodes which were partially used are having higher energy level than that of original DSR. The life time of these nodes are drastically improved and as

2012 2nd IEEE International Conference on Parallel, Distributed and Grid Computing

their remaining energy is more in ECDSR as compare to DSR, they can be used further for the data communication.

4) Network life time We have considered network life time as one of the crucial factors while considering perfonnance analysis. From Figure-4 it is found that the life time of the node has been improved significantly for all most all nodes, consequently the life time of the node has also increased in ECDSR. In DSR the whole network gets collapsed at 3 1.026 second whereas in ECDSR it last till 49.83 1 sec. So there is an 18.82 sec improvement in the network life time which in percentage it is 60.6 1 percentage of improvement.

5) Number of dropping nodes

The Figure-5 depicts that the number of dropping nodes has been reduced to half. In DSR the number of dropping packet observed as 4, where in ECDSR it is only 2. This is another crucial parameter that we often neglect while considering the performance parameter. When a network contains less number of dropping nodes then the reliability and the life time of the network becomes higher. As our method has the survival feature, hence it avoids the nodes to participate in active communication process and the nodes which have high residual battery power will be more suitable path for the communication which may be carried out without any interruption.

6) Throughput o{the network In order to make the DSR as an energy conscious DSR, we have used some of the extra functionalities and mechanism like energy saving and survival where all nodes are updating their status continuously. Due to this more number of control packets is flown from different paths from different nodes of the network. Due to this overuse of control packet there is a slightly degradation in the throughput as shown in Figure-6. However the variation of the throughput can be ignored as the difference is quite low.

IV. CONCLUSION AND FUTURE WORK

In this paper we have designed an energy conscious DSR where two new features are included with the basic DSR routing mechanism. ECDSR is not only turning out to be an energy conscious routing protocol but also proved itself as a better routing protocol than the traditional DSR. In all perfonnance metrics ECDSR has shown better results than DSR. Our method has improved the individual node life time and so the whole network life span with a 60.6 1 % of improvement with respect to traditional DSR. It is also found that the energy consumption per successful packet delivery is 0. 1 joule lesser than that of DSR. Packet delivery ration has been increased from 82.7 % to 85.3 %. Interestingly our proposed ECDSR reduces the number of dropping packet from 4 to 2 as ECDSR chooses its first path with the higher traffic time as the selection procedure is based on the residual power of the individual nodes. This paper has left many scopes for the researcher to extend this work. As we have not considered the mobility of the nodes, it may give some interesting results when the same method will be used for dynamic networks. We may carry out

789

this work with more number of nodes. So in dense network, can it perform in the same way is to be noticed. We have only taken a single source and destination, so with multiple sources and destinations with UDP connection can also be another variation of the experiments.

REFERENCES

[1] The C. Siva Ram Murthy, B. S. Manoj, "Ad Hoc Wireless

Networks Architecture and Protocols", 2nd ed, Pearson Education, 2005.

[2] E. Royer and C.-k. Toh, "A Review of Current Routing Protocols for Ad Hoc Moblie Wireless Networks," IEEE Personal Comm.Magazine, vol. 6, no. 2, Apr. 1999.

[3] C. Yu, B. Lee, H. Youn, "Energy Efficient Routing Protocols for Mobile Ad Hoc Networks". Wireless Communication and Mobile Computing, Wireless Com. Mob. Computing (2003).

[4] Chang J-H, tassiulus L., "Energy Conserving Routing in Wireless Ad Hoc networks", In Proceedings of IEEE FOCOM, March, 2000,pages 22-31.

[5] Stojmenovic I., Lin X., "Power Aware Localized Routing in Wireless Networks", IEEE transaction on parallel and Distributed Systems Vol. 12 issue.ll, November, 2001, pages 1122-1133.

[6] Doshi S., Bhandare s., Brown T.X., "An On-demand Minimum Energy Routing Protocol for Wireless Ad Hoc Networks", ACM SIGMOBTLE Mobile Computing and Communication Review, vol. 6, Issue 2, pages 50-66.

[7] Banergee S, Mishra A. "Minimum Energy Path for Reliable Communication in Multi-hop Wireless Network". In proceedings of the 3d ACM Annual Workshop on mobile Ad Hoc Networking and Computing (MobiHoc), June 2002, pages 146-156.

[8] Toh c.-K. "Maximum Battery Life Routing to Support

Ubiqutious Mobile Computing in Wireless Ad Hoc Networks", IEEE Communication Magazine, Vol. 39, issue. 6, June 2001, pages 138-147.

[9] Xu Y., Heidemann J., Estrin D. "Geography-informed Energy Conservation for Ad Hoc Routing", in Proceedings of 7th Annual international Conference on Mobile Computing and Networking (Mobicom), July 2001, pages 70-84.

[10] Feeney, L. M., Nilsson, M., "Investigating the Energy Consumption of Wireless Network Interface in an Ad Hoc Networking Environment", IEEE INFOCOM, vol. 3, April 2001, pages 1548-1557.

[11] A. Zhou, H. hassanein, "Load-Balanced Wireless Ad Hoc Routing", Proceedings of Canadian Conference on Electrical and Computer Engineering, vol. 2, 1157-1161 (2001).

[12] G. Chakrabarti, S. Kulkarni, "Load Balancing and Resource Reservation in Mobile Ad Hoc Networks", Ad Hoc Networks, Volume 4, issue 2, I March (2006).

[13] IRTEF Draft, The Dynamic Source Routing Protocol (DSR) for Mobile Ad Hoc Networks, available at: http://tools.ietf.org/html/rfc4 728 (2010).

[14] D.B Johnson, D.A Maltz, Y.-c. Hu, the Dynamic Source Routing Protocol for Mobile Ad Hoc Networks (DSR). A vailable from: http://www.ietf.org/internetdrafts/ draftietf­manet-dsr-l0.txt (2007).

[15] The network simulator (NS-), http:// www.isi.edu/nsnam/ns (2010).

[16] Fall. K. and Varadhan, K., "NS Notes and Documentation Technical report" University of California- Berkley, LBL, USC/lSI and Xerox P ARC.