Your SlideShare is downloading. ×
A Cross-Layer Based Multipath Routing Protocol To Improve QoS In Mobile Adhoc Networks
Upcoming SlideShare
Loading in...5
×

Thanks for flagging this SlideShare!

Oops! An error has occurred.

×

Introducing the official SlideShare app

Stunning, full-screen experience for iPhone and Android

Text the download link to your phone

Standard text messaging rates apply

A Cross-Layer Based Multipath Routing Protocol To Improve QoS In Mobile Adhoc Networks

386
views

Published on

In Mobile ad hoc networks, due to the high packet loss rates …

In Mobile ad hoc networks, due to the high packet loss rates
and frequent topological changes, the unbalanced transport
layer and reserved amount of traffic is carried out by the
network. In a QoS based routing metric for MANETs, it is
necessary to combine the minimum available bandwidth and
end-to-end delay along with the congestion around a link. In
this paper, a cross layer based multipath routing (CBMR)
protocol to improve QoS in mobile ad hoc networks to allot
weights to individual links, depending on the metrics link
quality, channel quality and end-to-end delay is developed. In
order to validate load balancing and interference between the
links using the same channel, the individual link weights are
integrated into a routing metric. Therefore, the weight value
helps the routing protocol to avoid the routing traffic through
the congested area hence the traffic is balanced and the
network capacity is improved. Then the proportion of traffic
to be routed to each neighbor is selected to execute routing
such that the weight of the node is a minimum. We also
propose an enhanced TCP congestion control mechanism for
wireless networks, based on a cross-layer scheme. By our
simulation results, the robustness of our protocol achieves
increased packet delivery ratio with reduced latency was
demonstrated.

Published in: Technology, Business

0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total Views
386
On Slideshare
0
From Embeds
0
Number of Embeds
0
Actions
Shares
0
Downloads
24
Comments
0
Likes
0
Embeds 0
No embeds

Report content
Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
No notes for slide

Transcript

  • 1. ACEEE International Journal on Network Security, Vol 1, No. 1, Jan 2010 A Cross-Layer Based Multipath Routing Protocol To Improve QoS In Mobile Adhoc Networks S.Venkatasubramanian1 Asst.Professor/Department of computer science & Engineering, Saranathan college of Engineering, Panjapur, Tiruchirapalli, India Pin: 620 012 veeyes@saranathan.ac.inAbstract of service (QoS). When the nodes communicate over wireless links then all the nodes should fight against theIn Mobile ad hoc networks, due to the high packet loss rates unpredictable character of the wireless channels andand frequent topological changes, the unbalanced transport intrusion from the additional transmitting nodes. Even iflayer and reserved amount of traffic is carried out by thenetwork. In a QoS based routing metric for MANETs, it is the user-required QoS in wireless ad hoc networks isnecessary to combine the minimum available bandwidth and attained, these factors make it as a challenging problem toend-to-end delay along with the congestion around a link. In use on the data throughput.this paper, a cross layer based multipath routing (CBMR)protocol to improve QoS in mobile ad hoc networks to allot Due to the mobility of the nodes, repeated route changesweights to individual links, depending on the metrics link cause difficulty in implementing ad hoc networks. Due toquality, channel quality and end-to-end delay is developed. In the high packet loss rates and regular topological changes,order to validate load balancing and interference between the the transport layer is not balanced and the inhibited amountlinks using the same channel, the individual link weights are of traffic is carried out by the network. The three famousintegrated into a routing metric. Therefore, the weight valuehelps the routing protocol to avoid the routing traffic through problems in ad hoc networks are: due to the intrusion andthe congested area hence the traffic is balanced and the movement of nodes, the reliability of the packet delivery isnetwork capacity is improved. Then the proportion of traffic not sufficient, incomplete bandwidth due to channelto be routed to each neighbor is selected to execute routing limitations and the inhibited node life span caused as ansuch that the weight of the node is a minimum. We also outcome of small battery size.propose an enhanced TCP congestion control mechanism forwireless networks, based on a cross-layer scheme. By our B. Problems of Adhoc Routing Protocolssimulation results, the robustness of our protocol achievesincreased packet delivery ratio with reduced latency was (i) An excellent scalable QoS architecture is required fordemonstrated. the huge dimension of ad hoc networks. An ad hoc network may not want to (or be able) to limit the numberIndex Terms: CBMR,QoS,Congestion,MANET of hosts concerned in the network. The potential for the collisions and contention increases when several nodes I .Introduction join as ad hoc network or the data traffic increases and due to this the protocols face the challenging task to route dataA. Quality of Service (QoS) in MANET packets and results in scalability. The scalability problems have been considered in the QoS in ad hoc networks in theAd hoc wireless network is a special case of wireless previous works.network which consists of backbone infrastructure whichis determined before. This can be made flexible and (ii) By the Network Interface Queues (IFQ) which areorganized quickly by using the features of the wireless ad implemented by AODV [1], TORA [2] and DSR [3], thehoc networks. However, using this property important packets which are ready to be transmitted are buffered.technological challenges are also created. There are many These packets are received by the network protocol stack.challenges including the issues of efficient routing, Generally, the maximum numbers of packets which aremedium access, power management, security and quality present in the IFQ are limited. In addition to this, a1 The Author is an part-time research scholar in the area of wireless adhoc maximum timeout policy for packets in the IFQ isnetworks at Anna University,Trichy,India and can be reached at implemented by the IFQ. Therefore any packet in919443144356 expectation of a route in the IFQ for a huge period may be discarded without any notification [4]. Hou and Tipper [5] 42 © 2010 ACEEE DOI: 01.ijns.01.01.09
  • 2. ACEEE International Journal on Network Security, Vol 1, No. 1, Jan 2010have stated that the edge over of the IFQ of congestednodes is one of the key motives for the rejection in The remainder of the paper is organized as follows. Inthroughput for congested networks running DSR. Section II some of the existing work in this area is reviewed. The system design and overview are presented(iii) The existing routing protocols faced a problem, called in the Section III. Section IV presents the details of the communication Gray Zone problem, which is a estimating the QoS based routing metrics. Section V major challenge [4]. Based on the control (RREQ) describes our proposed QoS based robust multipath routing packets, the detection and establishments of end to protocol in detail. The experimental results are presented end routes are carried out in most of the routing in Section VI. protocols like AODV. However, these control packets and data packets have different properties. For example, when RREQ packets are compared with the II. RELATED WORK data packets, it shows that RREQ packets are smaller and are transmitted at lower bit rates. The Andre Schumacher, et. al. (2006), have approached the bidirectional RREQ packets can be transmitted or problem of load balancing for wireless multihop networks received by the nodes which are at the distance. On by distributed optimization using an approximation the other hand, as a result of the above mentioned algorithm for minimizing the maximum network property, the high bit-rate data packets cannot be sent congestion and implement it as a modification of the DSR or received. Due to the resultant “fragile” link which routing protocol. The algorithm was based on shortest-path triggers link repairs, high control overhead for computations that are integrated into the DSR route protocols like AODV is obtained. discovery and maintenance process. Therefore, it does not rely on the dissemination of global information within the(iv) The essential element of any QoS-enabled routing entire network [7].Kaixin Xu, et. al., (2003) have proposed protocol is the end-to-end delay estimation. In order to a Scalable QoS architecture suitable for large scale mobile calculate the end-to-end delay, the existing protocols ad hoc networks with the help of the scalable LANMAR establish the time taken to route RREQ and RREP routing protocol. They have also introduced the mobile packets along the specified path. However, RREQ and backbone network structure to enhance the network RREP packets are impossible to meet the similar performance [8].Jianbo Xue, et. al., (2003) have proposed levels of traffic delay and loss like data packets which a QoS framework for MANETs Adaptive Reservation and are considerably different from usual packets. Pre-allocation Protocol (ASAP). By using two signaling messages, ASAP provides fast and efficient QoS supportIn order to monitor the movement of the node and that a while maintaining adaptation flexibility and minimizingroute is about to break, the communication is based on the wasted reservations [9].received power of the forwarding node. Therefore, thequality of the route is measured. Based on the measured Jangeun Jun and Mihail L. Sichitiu, (2003) have exposed aquality of the route, the link breakage can be assumed. A significant fairness problem existent practically in allsmaller end-to-end delay and a larger throughput to a host wireless multihop networks. They showed that a networkcan be obtained by using a proactive fault tolerant routing layer solution can restore fairness at the expense ofwith QoS aware multipath route discovery. bandwidth efficiency[10].Michael Gerharz, et. al., (2003) have have outlined that it is a very complex task to reserveIn this paper, a cross layer based multipath routing or even measure available bandwidth in wireless multihop(CBMR) protocol to improve QOS in mobile ad hoc networks. They have introduced and evaluated the DLitenetworks intended for mobile ad hoc networks has been algorithm, an approach to service differentiation in ad hocdeveloped. In this paper, the facility to determine multiple networks. It applies a fair queuing scheme with separateroutes to a host and switch between them to expand the queues for each service class. Late packets of delaydefinition of AOMDV [6] has been employed. Enabling a constrained classes were dropped in intermediate routers.QoS constrained route from source to destination is one of [11].the objectives of the routing protocol. The route chosen bythe protocol must send packets with minimum bandwidth Duc A. Tran and Harish Raghavendra, (2006) haveand end-to-end latency, in particular. The protocol should proposed CRP [12], a congestion-adaptive routing protocolsatisfy the above constraints and also select the most for MANETs. CRP tried to prevent congestion fromrobust among all possible candidate routes.Different from occurring in the first place, rather than dealing with itwired networks, TCP over wireless networks needs to be reactively. Xiaoqin Chen et al.,(2007) [13] have proposedcross-layer designed which depends on information of a congestion- aware routing protocol for mobile ad hocMAC. However, it is difficult to apply existing cross-layer networks which uses a metric incorporating data-rate,design into hybrid networks. In this paper, we have also MAC overhead, and buffer delay to combat congestion.proposed an enhanced TCP congestion control mechanismfor wireless networks, based on a cross-layer scheme. 43© 2010 ACEEEDOI: 01.ijns.01.01.09
  • 3. ACEEE International Journal on Network Security, Vol 1, No. 1, Jan 2010Ming Yu et al.,(2007) [14] have proposed a link routing traffic through congested area. Subsequently, theavailability-based QoS-aware (LABQ) routing protocol for selection of the proportion of traffic to be routed to eachmobile ad hoc networks based on mobility prediction and neighbor is made to perform routing such that the weightlink quality measurement, in addition to energy of the node is a possible minimum.consumption estimate..Hongxia Sun et al [15], proposed anadaptive QoS routing scheme which is supported by cross-layer cooperation in ad hoc networks. By considering the IV. ESTIMATING QOS-BASED ROUTING METRICSinfluence of node mobility and lower-layer linkperformance, the cross-layer mechanism provides up-to- A. Estimating Link Qualitydate local QoS information for the adaptive routing Each node in the network estimates its quality of links withalgorithm. Based on the current network status, the its one-hop neighbors. If Nq is the number of HELLOmultiple QoS requirements are satisfied by adaptively packets received during a time window Twin and Pq is theusing forward error correction and multipath routing percentage of HELLO packets received in the last rmechanisms. Al-khwildi et al [16], have proposed a novel seconds, then the link quality Lq is measured asrouting technique called Adaptive Link-Weight (ALW)routing protocol which selects an optimum route on the Lq = δ. Pq + (1 – δ) . Nq (1)basis of available bandwidth, low delay and long routelifetime. Those technique adapts a cross-layer framework The estimated link quality is maintained by each node inwhere the ALW is included with application and physical its NT. The average link quality of all the links across thelayer. path P, gives the route quality Rq of the path. RREQ packets of the reverse path and RREP packets of the III. SYSTEM DESIGN AND PROTOCOL OVERVIEW forward path accumulate the estimated Lq values.The value δ is application specific.We devise the model by considering a mobile ad hocnetwork in which each node utilizes IEEE 802.11 DCF for B. Estimating End to End Delaymedium access control (MAC). At the application layer,intra-flow interference occurs for the same flow which is There is a significant variation between the end-to-endtransmitted on different wireless links. But it has been delay reported by RREQ-RREP measurements and thepresumed as a single data flow. delay experienced by actual data packets. We address thisA QOS based multipath routing protocol intended for issue by introducing a DUMMY-RREP phase during routemobile ad hoc networks is put forwarded. In this paper, we discovery. The source saves the RREP packets it receivesemployed the facility to determine multiple routes to a host in a RREP TABLE and then acquires the RREP for a routeand switch between them to expand the definition of from this table to send a stream of DUMMY data packetsAOMDV [6].Enabling a QoS constrained route from along the path traversed by this RREP. DUMMY packetssource to destination is the objective of the routing efficiently imitate real data packets on a particular pathprotocol. The route chosen by the protocol must send owing to the same size, priority and data rate as real datapackets with minimum bandwidth and end-to-end latency, packets. H is the hop count reported by the RREP. Thewithout facing congestion. The protocol should satisfy the number of packets comprised in every stream is 2H. Theabove constraints and also select the most robust among all destination computes the average delay Davg of allpossible candidate routes. DUMMY packets received, which is sent through a RREP to the source. The source selects this route and sends dataA QoS-based routing metric for MANETs should packets only when the average delay reported by thisincorporate minimum available bandwidth and end-to-end RREP is inside the bound requested by the application.latency along with congestion around a link. Congestion is The source performs a linear back-off and sends therelated to channel quality, which depends on the MAC DUMMY stream on a different route selected from itsaccess contention and channel reliability. So our algorithm RREP TABLE when the delay exceeds the required limit.should rely on the following metrics to allocate weights to C. Estimating Channel Qualityindividual links. • End-to-End Delay The channel quality can be represented by both the channel • Channel Quality occupancy and channel reliability. • Link Quality C.1. Channel OccupationThe individual link weights are combined into a routing In this network, we consider IEEE 802.11 MAC with themetric to validate the load balancing and interference distributed coordination function (DCF). It has the packetbetween links using the same channel. Consequently, the sequence as request-to-send (RTS), clear-to-send (CTS),traffic is balanced and the network capacity is improved as data and acknowledgement (ACK). The amount of timethe weight value assists the routing protocol to evade between the receipt of one packet and the transmission of 44© 2010 ACEEEDOI: 01.ijns.01.01.09
  • 4. ACEEE International Journal on Network Security, Vol 1, No. 1, Jan 2010the next is called a short inter frame space (SIFS). Then SSRC with an initial value of 0, when it transmits the RTSthe channel occupation due to MAC contention will be frame successfully. By this way, when a node starts to transmit a new RTS frame, SSRC will be the initial value Cocc = tRTS + tCTS + 3tSIFS + tacc (2) of 0, in spite of the initial RTS which was transmitted successfully or discarded. Obviously, the node would notWhere tRTS and tCTS are the time consumed on RTS and know the RTS retry count of the last frame for its resettingCTS, respectively and tSIFS is the SIFS period. tacc is the which results that its channel state cannot be known.time taken due to access contention. The channeloccupation is mainly dependent upon the medium access To solve this problem, the value of SSRC is recorded whencontention, and the number of packet collisions. That is, each RTS frame is retransmitted. The packet to beCocc is strongly related to the congestion around a given transmitted will be marked as “1” in its congestionnode. Cocc can become relatively large if congestion is experience bit when it exceeds the given the threshold “2”,incurred and not controlled, and it can dramatically which shows an imminent congestion in ECN mechanism.decrease the capacity of a congested link.Therefore, in the design of a congestion-aware metric for VI. QOS BASED ROBUST MULTIPATH ROUTINGnetworks, the access contention should be considered to PROTOCOLmore accurately indicate channel capacity. For establishing multiple disjoint paths, we adapt the ideaC.2. Channel Reliability from the Adhoc On-Demand Multipath Distance VectorThe channel reliability also affects the packet transmission Routing (AOMDV) [6]. The multiple paths are computedin MANETs and factors such as interference and fading in during the route discovery.the channel may lead to occurrence of packet losses. In802.11 DCF, several failed retransmissions lead to packets We now introduce the weight metric W which assigns abeing dropped. In addition to performance deterioration cost to each link in the network. The weight W combinesdue to congestion with respect to packet losses, higher the link quality Lq , channel quality Cocc and the averagepacket retransmissions involved owing to augmented delay Davg , to select maximum throughput paths, avoidingpacket losses lead to more congestion. A successful the most congested links.For an intermediate node i withtransmission of a packet on an unreliable links would established transmission with several of its neighbours, theconsume more time on MAC overhead. W for the link from node i to a particular neighboring node is given by V. CROSS-LAYER DESIGN OF TCP W = Lq + Cocc + Davg (3) A. Route RequestCurrently, the main method of congestion control overwired network is an optimization framework based on During the route discovery phase of the protocol, eachutility functions Moreover this method is not applicable intermediate node uses an admission control scheme toto the wireless networks. This is because due to the check whether the flow can be accepted or not. If accepted,immanent interference of wireless links the metrics like a Flow Table (FT) entry for that particular flow is created.queue length on routers, sending rate of links cannot The FT contains the fields Source (Src), Destination (Dst),characterize the congestion price. Reserved Bandwidth (BWres), Minimum bandwidthIn this paper, according to the RTS retry counts, the ECN (BWmin). Each node collects the bandwidth reserved at itsscheme will mark packets. IEEE 802.11 is the standard one hop neighbors (piggybacked on periodic HELLOprotocol for wireless adhoc networks. It provides two packets) and stores it in its Neighbor Table (NT) .Theoptions for accessing namely the basic scheme and Neighbor Table contains fields Destination (Dst), ReservedRTS/CTS scheme. In this paper only this RTS/CTS Bandwidth (BWres), No. of Hello Packets (No Hello).scheme is considered because it is used more in ad hocnetworks than the basic scheme. But our design can also be Consider the scenarioapplied to the basic scheme. Let us consider the routeIn RTS/CTS scheme, in order to record the retransmitcount of RTS frame in the current node, each node S -- R1 – R2 – R3 – Dmaintains a variable SSRC. When a node fails to transmitRTS frame, then it will retransmit this RTS frame and To initiate QoS-aware routing discovery, the source host Sincreases SSRC by 1 at the same time. The sender will sends a RREQ. When the intermediate host R1 receivesreport a link failure notification to the network layer if the the RREQ packet, it first estimates all the metrics asvalue of the SSRC reaches 7 and discards the RTS frame described in the previous section.and the data frame to be transmitted together and resetsSSRC to an initial value of 0. Also the node refreshes the 45© 2010 ACEEEDOI: 01.ijns.01.01.09
  • 5. ACEEE International Journal on Network Security, Vol 1, No. 1, Jan 2010The host R1 then calculates its weight WR1 using (3). Table1: Simulation Parameters WR1 No. of Nodes 25,50,75 and 100 RREQR1 ====== R2 Area Size 1000 X 1000R2 then calculates its weight WR2 in the same way and Mac 802.11adds it to the weight of R1. R2 then forward the RREQ Radio Range 250mpacket with this added weight. Simulation Time 100 sec Traffic Source CBR Packet Size 512 WR1+ WR2 Mobility Model Random Way Point RREQR2 ============= R3 Speed 5m/s to 20m/sFinally the RREQ reaches the destination node D with the Pause time 5ssum of node weights B. Performance Metrics WR1+ WR2 + WR3 RREQR3 ================ D We compare our CBMR protocol with the AOMDV [6], CARM [13] and LABQ [14] protocol. We evaluate mainlyB. Route Reply the performance according to the following metrics, by varying the nodes as 25, 50, 75 and 100.The Destination node D send the route reply packet RREPalong with the total node weight to the immediate Control overhead: The control overhead is defined as theupstream node R3 . total number of routing control packets normalized by the total number of received data packets. WR1+ WR2 + WR3 RREP ================ R3 Average end-to-end delay: The end-to-end-delay is averaged over all surviving data packets from the sourcesNow R3 calculates its cost C based on the information to the destinations.from RREP as Average Packet Delivery Ratio: It is the ratio of the No.CR3 = (WR1+ WR2 +WR3) - (WR1+ WR2) (4) of packets received successfully and the total no. of packets sentBy proceeding in the same way, all the intermediate hostscalculate its cost. C. ResultsOn receiving the RREP from all the routes, the source C.1. Varying No. Of Nodesselects the route with minimum cost value. In the first experiment, the performance of the protocols isThe simulation results in the next section showed that measured by varying the no. of nodes as 25, 50, 75 andthese calculations did not increase the overhead 100. From Figure 1, we can see that the packet deliverysignificantly. ratio for CBMR increases, when compared to AOMDV, CARM and LABQ, since it utilizes robust links. VII. SIMULATION RESULTS From Figure 2, it is evident that the average end-to-end delay of the proposed CBMR protocol is less whenA. Simulation Model and Parameters compared to the AOMDV, CARM and LABQ protocol.NS2 simulator is used to simulate our proposed protocol. Figure 3 shows that the control overhead of the protocols.In our simulation, the channel capacity of mobile hosts is Since CBMR make use of HELLO packets for costset to the same value: 2 Mbps. We use the distributed estimation, the values are considerably less in CBMRcoordination function (DCF) of IEEE 802.11 for wireless when compared with AOMDV, CARM and LABQ.LANs as the MAC layer protocol. It has the functionalityto notify the network layer about link breakage.In our simulation, each node moves independently with thesame average speed is assumed. The simulated traffic isConstant Bit Rate (CBR). Our simulation settings andparameters are summarized in table 1 46© 2010 ACEEEDOI: 01.ijns.01.01.09
  • 6. ACEEE International Journal on Network Security, Vol 1, No. 1, Jan 2010 Packet Delivery Ratio Packet Delivery Ratio 100 120 100 CBMR 80 CBMR 80 ratio(%) ratio(% ) AOMDV 60 AOMDV 60 CARM 40 CARM 40 20 LABQ 20 LABQ 0 0 250 500 750 1000 25 50 75 100 Rate Nodes Fig.4: Rate Vs Delivery Ratio Fig 1: Nodes Vs Delivery Ratio From Figure 5, we can see that the average end-to-end End-to-End Delay delay of the proposed CBMR protocol is less when compared to AOMDV, CARM and LABQ. 0.8 Figure 6 shows the control overhead of the protocols. 0.6 CBMR Since CBMR make use of HELLO packets for cost Delay(s) 0.4 AOMDV estimation, the values are considerably less in CBMR CARM when compared with AOMDV, CARM and LABQ. 0.2 LABQ 0 25 50 75 100 Nodes End-to-End Delay 0.6 Fig. 2: Nodes Vs Delay 0.5 CBMR 0.4 Delay(s) AOMDV 0.3 Control Overhead CARM 0.2 0.1 LABQ 1.2 0 1 CBMR 250 500 750 1000 0.8 AOMDV Rate ratio 0.6 0.4 CARM 0.2 LABQ 0 Fig. 5: Rate Vs Delay 25 50 75 100 Nodes Control Overhead 1.2 1 Fig. 3: Nodes Vs Overhead CBMR 0.8 ratio(%) AOMDV 0.6B. Varying the Transmission Rate 0.4 CARMIn the second experiment, we measure the performance of 0.2 LABQthe protocols by varying the transmission rate as 0250,500,750 and 1000 Kb. 250 500 750 1000 RateFrom Figure 4, we can see that the packet delivery ratio forCBMR is more, when compared to AOMDV, CARM andLABQ since it utilizes robust links. Fig 6: Rate Vs Overhead 47© 2010 ACEEEDOI: 01.ijns.01.01.09
  • 7. ACEEE International Journal on Network Security, Vol 1, No. 1, Jan 2010 CONCLUSION Communications, vol. 2, pp: 1018- 1023, 13- 16 October 2003, Doi: 10.1109/MILCOM.2003.1290305.In a QoS based routing metric for MANETs, it is necessary [9] Jianbo Xue, Patrick Stuedi and Gustavo Alonso, "ASAP: Anto combine the minimum available bandwidth and end-to- Adaptive QoS Protocol for Mobile Ad Hoc Networks",end delay along with the congestion around a link. MAC IEEE Proceedings on Personal, Indoor and Mobile Radio Communications, vol. 3, pp: 2616- 2620, 7- 10 Septemberaccess contention and channel reliability induces 2003, Doi: 10.1109/PIMRC.2003.1259201.congestion that is related to channel quality. In this paper [10] Jangeun Jun and Mihail L. Sichitiu, "Fairness and QoS inwe have developed a cross layer based multipath routing Multihop Wireless Networks", in proceedings of 58th IEEE(CBMR) protocol to improve QoS in mobile ad hoc Conference on Vehicular Technology, vol. 5, pp: 2936-networks to allot weights to individual links, depending on 2940, 6- 9 October 2003, Doi:the metrics link quality, channel quality and end-to-end 10.1109/VETECF.2003.1286161.delay. The weight value helps the routing protocol to avoid [11] Michael Gerharz, Christian de Waal, Matthias Frank andthe routing traffic through the congested area and hence Paul James, "A Practical View on Quality-of-Servicethe traffic is balanced and the network capacity is Support in Wireless Ad Hoc Networks", Proc. of the 3rd IEEE Workshop on Applications and Services in Wirelessimproved. Then the proportion of traffic to be routed to Networks (ASWN), pp: 185-196, Berne, Switzerland, Julyeach neighbor is selected to execute routing such that the 2003.weight of the node is a minimum. We have also proposed an [12] Duc A. Tran and Harish Raghavendra, "Congestionenhanced TCP congestion control mechanism for wireless Adaptive Routing in Mobile Ad Hoc Networks", IEEEnetworks, based on a cross-layer scheme. By simulation Transactions on Parallel and Distributed Systems, Vol. 17,results, we have demonstrated that the robustness of our no. 11, November 2006.protocol achieves increased packet delivery ratio with [13] Xiaoqin Chen, Haley M. Jones, A .D .S. Jayalath,reduced latency. The security challenges of our proposed "Congestion-Aware Routing Protocol for Mobile Ad Hocprotocol will be the subject of our future work. Networks", in proceedings of 6th IEEE Conference on Vehicular Technology, pp: 21- 25, 30 September- 3 October, Baltimore, 2007, Doi: 10.1109/VETECF.2007.21. REFERENCES [14] Ming Yu , Aniket Malvankar , Wei Su , Simon Y. Foo , "A link availability-based QoS-aware routing protocol for[1] Charles E. Perkins and Elizabeth M. Royer, "Ad-hoc On- mobile ad hoc sensor networks", Journal on Computer Demand Distance Vector Routing", in proceedings of 2nd Communications, vol. 30, no. 18, pp: 3823- 3831, December workshop on Mobile Computing Systems and Applications, 2007. pp: 90- 100, 25-26 Feb., New Orleans, LA, USA, 1999, [15] Hongxia Sun, Herman D. Hughes, “Adaptive QoS Routing Doi: 10.1109/MCSA.1999.749281. by Cross-Layer Cooperation in Ad Hoc Networks”,[2] Vincent D. Park and M. Scott Corson, "A Highly Adaptive EURASIP Journal on Wireless Communications and Distributed Routing Algorithm for Mobile Wireless Networking 2005:5, 661–671. Networks", in Proc. of IEEE INFOCOM97, vol. 3, pp: [16] A.N. Al-Khwildi, S. Khan, K.K. Loo, H.S. Al-Raweshidy, 1405-1413, 7-12 April, Kobe, Japan, 1997, Doi: “Adaptive Link-Weight Routing Protocol using Cross-Layer 10.1109/INFCOM.1997.631180. Communication for MANET”, WSEAS TRANSACTIONS on[3] David B. Johnson and David A. Maltz, "Dynamic Source COMMUNICATIONS Manuscript received May 16, 2007; Routing in Ad Hoc Wireless Networks", in proceedings of revised Aug. 20, 2007. International Series in Engineering and Computer Science, vol. 353, pp: 153- 181, 1996.[4] Asad Amir Pirzada and Chris McDonald, "Inherent Robustness of Reactive Routing Protocols against Selfish Attacks", in proceedings of International Workshop on Wireless Ad-hoc Networks, London, England, 2005.[5] Xiaobing Hou and David Tipper, "Impact of Failures on Routing in Mobile Ad Hoc Networks Using DSR", in Web Protocols White Papers, University of Pittsburgh, November 2002.[6] Mahesh K. Marina and Samir R. Das, "On- Demand Multipath Distance Vector Routing in Ad Hoc Networks", in proceedings of 9th IEEE International Conference on Network Protocols, pp: 14- 23, 11- 14 November 2001.[7] Andre Schumacher, Harri Haanpaa, Satu Elisa Schaeffer, and Pekka Orponen, "Load Balancing by Distributed Optimization in Ad Hoc Networks", in proceedings of 2nd International Conference on Mobile Ad-hoc and Sensor Networks, vol. 4325, pp: 874- 885, Berlin- Heidelberg, 2006.[8] Kaixin Xu, Ken Tang, Rajive Bagrodia, Mario Gerla and Michael Bereschinsky, "Adaptive Bandwidth Management and QoS Provisioning in Large Scale Ad Hoc Networks", in proceedings of IEEE Conference on Military 48© 2010 ACEEEDOI: 01.ijns.01.01.09