Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.



Published on

International Journal of Engineering Research and Applications (IJERA) is an open access online peer reviewed international journal that publishes research and review articles in the fields of Computer Science, Neural Networks, Electrical Engineering, Software Engineering, Information Technology, Mechanical Engineering, Chemical Engineering, Plastic Engineering, Food Technology, Textile Engineering, Nano Technology & science, Power Electronics, Electronics & Communication Engineering, Computational mathematics, Image processing, Civil Engineering, Structural Engineering, Environmental Engineering, VLSI Testing & Low Power VLSI Design etc.

Published in: Technology, Business
  • Be the first to comment

  • Be the first to like this


  1. 1. B.Sindhu, V.Bharathi, N.Karthikeyan / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.092-09992 | P a g eAutonomous Reconfiguration for Wireless Mesh NetworkB.Sindhu*, V.Bharathi**,N.Karthikeyan***Department of Electronics and Communication Engineering,Sri Manakula Vinayagar Engineering College, MadagadipetABSTRACTThe demands for the network usage areincreasing day by day. During their lifetime,multihop wireless mesh networks (WMNs)experience frequent link failures caused bychannel interference, dynamic obstacles, and/orapplications’ bandwidth demands, which maydegrade the system performance. This paperpresents an autonomous networkreconfiguration system (ANRS) that enables amultiradio WMN to autonomously recover fromlocal link failures to preserve networkperformance. Using ANRS necessary changesare made in local and radio channel assignmentsfor failure recovery. Next, based on the thus-generated configuration changes, the systemcooperatively reconfigures network settingsamong local mesh routers. ANRS has beenimplemented extensively on our IEEE 802.11-based WMN through ns2-based simulation. Ourevaluation results show that ANRS outperformsexisting failure-recovery schemes in improvingchannel-efficiency and in the ability of meetingthe applications’ bandwidth demands.Keywords-IEEE 802.11, multiradio wirelessmesh networks (mr-WMNs), Routing, self-reconfigurable networks, wireless link failures.1. INTRODUCTIONWireless mesh networks are boon to thewireless architecture. It supports larger applicationsand it provides several benefits to users such as, nocabling cost, automatic connection to all nodes,network flexibility, ease of installation and it alsodiscovers new routes automatically. These wirelessmesh networks are not stand alone it is compatibleand interoperable with other wireless networks. Itprovides greater range of data transfer rates.Wireless mesh networks are preferable compared tothe adhoc networks for the easy of networkmaintanance, robustness etc. Wireless meshnetworks[1] often consist of mesh clients, meshrouters and gateways. The mesh clients are oftenlaptops, cell phones and other wireless deviceswhile the mesh routers forward traffic to and fromthe gateways which may but need not connect tothe Internet.The coverage area of the radio nodesworking as a single network is sometimes called amesh cloud. Access to this mesh cloud is dependenton the radio nodes working in harmony with eachother to create a radio network. A mesh network isreliable and offers redundancy. When one node canno longer operate, the rest of the nodes can stillcommunicate with each other, directly or throughone or more intermediate nodes. A wireless meshnetwork can be seen as a special type of wirelessad-hoc network. A wireless mesh network often hasa more planned configuration, and may bedeployed to provide dynamic and cost effectiveconnectivity over a certain geographic area.Fig.1 wireless mesh networkWireless mesh networks (WMNs) arebeing developed actively and deployed widely for avariety of applications, such as public safety,environment monitoring, and citywide wirelessInternet services. They have also been evolving invarious forms (e.g., using multiradio /channel[2]systems to meet the increasing capacity demandsby the above-mentioned and other emergingapplications. However, due to heterogeneous andfluctuating wireless link conditions preserving therequired performance of such WMNs is still achallenging problem.For example, some links of a WMN mayexperience significant channel interference fromother coexisting wireless networks. Some parts ofnetworks might not be able to meet increasingbandwidth demands from new mobile users andapplications. Links in a certain area (e.g., a hospitalor police station) might not be able to use somefrequency channels because of spectrum etiquetteor regulation. Wireless mesh networks does notprovide centralized trusted architecture too, todistribute the public keys. Here in the wirelessmesh network, mesh clients should contain powerefficient protocols. Mesh routers in the wireless
  2. 2. B.Sindhu, V.Bharathi, N.Karthikeyan / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.092-09993 | P a g emesh networks perform dedicated routing andconfigurations.2. RELATED WORKEven though many solutions for WMNs torecover from wireless link failures have beenproposed, they still have several limitations asfollows. First, resource-allocation algorithms canprovide (theoretical) guidelines for initial networkresource planning. However, even though theirapproach provides a comprehensive and optimalnetwork configuration plan, they often require―global‖ configuration changes, which areundesirable in case of frequent local link failures.Next, a greedy channel-assignment algorithm canreduce the requirement of network changes bychanging settings of only the faulty link(s).However, this greedy change might not be able torealize full improvements, which can only beachieved by considering configurations ofneighbouring mesh routers in addition to the faultylink(s). Third, fault-tolerant routing protocols, suchas local rerouting or multipath routing can beadopted to use network-level path diversity foravoiding the faulty links. However, they rely ondetour paths or redundant transmissions, whichmay require more network resources than link-levelnetwork reconfiguration.2.1 CHANNELASSIGNMENTALGORITHMChannel assignment and schedulingalgorithms provide holistic guidelines, such asthroughput bounds and schedulability for channelassignment during a network deployment stage.However, the algorithms do not consider the degreeof configuration changes from previous networksettings, and hence they often require globalnetwork changes to meet all the constraints. Eventhough these algorithms are suitable for static orperiodic network planning, they may cause networkservice disruption, and thus are unsuitable fordynamic network reconfiguration that has to dealwith frequent local link failures.The greedy channel-assignment algorithm, whichconsiders only local areas in channel assignments,might do better in reducing the scope of networkchanges than the channel assignment algorithms.However, this approach still suffers from the rippleeffect, in which one local change triggers thechange of additional network settings atneighboring nodes due to association dependencyamong neighboring radios. This undesired effectmight be avoided by transforming a mesh topologyinto a tree topology, but this transformation reducesnetwork connectivity as well as path diversityamong mesh nodes.Interference-aware channel assignmentalgorithms[3] can minimize interference byassigning orthogonal channels as closely aspossible geographically. While this approach canimprove overall network capacity by usingadditional channels, the algorithm could furtherimprove. These algorithms may require globalnetwork configuration changes from changing localQoS demands, thus causing network disruption.3 PROPOSED SYSTEMTo overcome the above limitations, wepropose an autonomous network reconfigurationsystem (ANRS) that allows a multiradio WMN(mr-WMN) to autonomously reconfigure its localnetwork settings—channel, radio, and routeassignment— for real-time recovery from linkfailures[5]. In its core, ANRS is equipped with areconfiguration planning algorithm that identifieslocal configuration changes for the recovery whileminimizing changes of healthy network settings.Briefly, ANRS first searches for feasiblelocal configuration changes available around afaulty area, based on current channel and radioassociations. Then, by imposing current networksettings as constraints, ANRS identifiesreconfiguration plans that require the minimumnumber of changes for the healthy network settings.Next, ANRS also includes a monitoring protocolthat enables a WMN to perform real-time failurerecovery in conjunction with the planningalgorithm. The accurate link-quality informationfrom the monitoring protocol is used to identifynetwork changes that satisfy applications’ new QoSdemands or that avoid propagation of QoS failuresto neighboring links (or ―ripple effects‖). Ourevaluation results show that ANRS outperformsexisting failure-recovery methods, such as static orgreedy channel assignments, and local rerouting[4].First, ANRS’s planning algorithmeffectively identifies reconfiguration plans thatmaximally satisfy the applications’ QoS demands.Next, ANRS avoids the ripple effect via QoS-awarereconfiguration planning, unlike the greedyapproach. Third, ANRS’s local reconfigurationimproves network throughput and channelefficiency by more than 26% and 92%,respectively, over the local rerouting scheme. Therouting messages are being transmitted from thesource node to the destination in order to find theleast cost path. These routing messages shouldatleast carry one ticket regularly.3.1 Overall modules of ANRSThe overall modules of ANRS system are,i. Network constructionii. Link-Stateiii. Group organizeriv. Failure detectorv. Gateway plannera. Plan generatorb. QOS filtervi. benefit filtervii. Optimal Analyser
  3. 3. B.Sindhu, V.Bharathi, N.Karthikeyan / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.092-09994 | P a g eFig 2: ANRS software architectureA. Network ConstructionIn this module, construct the network ascluster form. It store the node name, data rate and italso maintain the Information about theneighbouring nodes .Gateway also involved in it. Itwill be make detection of route as better.B. Link StateEvery node monitors the quality of itsoutgoing wireless links. The node containsinformation about the incoming and outgoingtraffic. It maintains the information about theneighbouring nodes. It measures wireless linkconditions via a hybrid link quality measurementtechnique.C. Group OrganizerIt forms a local group among meshrouters. Each router has a specific set of locationsfrom which it can accept data, and a specific set oflocations to which it can send data. Mesh routerswork by continuously monitoring network activityand maintaining lists of other devices in theirvicinity.D. Failure DetectorIt interacts with an network monitor in thedevice driver and maintains an up-to-date link-statetable. Network monitor, monitors link-quality andextensible to support as many multiple radios aspossible.E. Gateway PlannerNetwork Planner: It generates neededreconfiguration plans only in a gateway node.QoS Planner: ARS applies strict constraintto identify a reconfiguration plan that satisfies theQoS demands and that improves networkutilization most.Benefit Filter: It identifies, whichreconfiguration plans are suitable to reachdestination.F. Optimal plannerIt identifies, which reconfiguration plan havingthe shortest path to reach destination. Identifiesreconfiguration plans that require the minimumnumber of changes for the healthy network settings.3.2 Planning for localized reconfigurationThe core function of ANRS is to systematicallygenerate localized reconfiguration plans. Areconfiguration plan is defined as a set of links’configuration changes (e.g., channel switch, linkassociation) necessary for a network to recoverfrom a link(s) failure on a channel, and there areusually multiple reconfiguration plans for each linkfailure.Fig 3: Localized reconfiguration planning inANRSExisting channel-assignment andscheduling algorithms seek ―optimal‖ solutions byconsidering tight QoS constraints on all links, thusrequiring a large configuration space to be searchedand hence making the planning often an NP-complete problem.In addition, change in a link’s requirementmay lead to completely different networkconfigurations. By contrast, ANRS systematicallygenerates reconfiguration plans that localizenetwork changes by dividing the reconfigurationplanning into three processes—feasibility, QoSsatisfiability, and optimality—and applyingdifferent levels of constraints.As depicted in Fig. 3, ANRS first appliesconnectivity constraints to generate a set of feasiblereconfiguration plans that enumerate feasiblechannel, link, and route changes around the faultyareas, given connectivity and link-failureconstraints.Then, within the set, ANRS applies strictconstraints (i.e., QoS and network utilization) toidentify a reconfiguration plan that satisfies theQoS demands and that improves networkutilization most. It also generates feasible plan forreconfiguration and also provide QoS satisfiabilityevalution for better results.3.2.1 Feasible Plan Generation:Generating feasible plans is essentially tosearch all legitimate changes in links’configurations and their combinations around thefaulty area. Given multiple radios, channels, androutes, ANRS identifies feasible changes that helpavoid a local link failure but maintain existingnetwork connectivity as much as possible.However, in generating such plans,ANRS has to address the following challenges.
  4. 4. B.Sindhu, V.Bharathi, N.Karthikeyan / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.092-09995 | P a g e1) Avoiding a faulty channel:ANRS first has to ensure that the faulty link needsto be fixed via reconfiguration. To this end, ANRSconsiders three primitive link changes Specifically,to fix a faulty link(s), ANRS can use: A) achannel-switch where both end-radios of link ABcan simultaneously change their tuned channel;B)a radio-switch where one radio in node A canswitch its channel and associate with another radioin node B; C) a route-switch where all traffic overthe faulty link can use a detour path instead of thefaulty link.TABLE 1Link changes in ANRS2) Maintaining network connectivity andutilization:While avoiding the use of the faultychannel, ANRS needs to maintain connectivitywith the full utilization of radio resources. Becauseeach radio can associate itself with multipleneighboring nodes, a change in one link triggersother neighboring links to change their settings. Tocoordinate such propagation, ANRS takes a two-step approach.ANRS first generates feasible changes ofeach link using the primitives, and then combines aset of feasible changes that enable a network tomaintain its own connectivity. Furthermore, for thecombination, ARS maximizes the usage of networkresources by making each radio of a mesh nodeassociate itself with at least one link and byavoiding the use of same (redundant) channelamong radios in one node.3) Controlling the scope of reconfigurationchanges:ANRS has to limit network changes aslocal as possible, but at the same time it needs tofind a locally optimal solution by considering morenetwork changes or scope. To make this tradeoff,ANRS uses a -hop reconfiguration parameter.Starting from a faulty link(s), ANRS considers linkchanges within the first hops and generates feasibleplans. If ANRS cannot find a local solution, itincreases the number of hops so that ANRS mayexplore a broad range of link changes. Thus, thetotal number of reconfiguration changes isdetermined on the basis of existing configurationsaround the faulty area as well as the value of k .3.3 NEED FOR SELF-RECONFIGURABILITYMaintaining the performance of WMNs inthe face of dynamic link failures remains achallenging problem. However, such failures canbe withstood (hence maintaining the requiredperformance) by enabling mr-WMNs toautonomously reconfigure channels and radio1assignments, as in the following examples.i) Recovering from link-quality degradation:The quality of wireless links in WMNscan degrade (i.e., link-quality failure) due to severeinterference from other collocated wirelessnetworks. For example, Bluetooth, cordless phones,and other coexisting wireless networks operatingon the same or adjacent channels cause significantand varying degrees of losses or collisions inpacket transmissions. By switching the tunedchannel of a link to other interference-freechannels, local links can recover from such a linkfailure.ii) Satisfying dynamic QoS demands:Links in some areas may not be able toaccommodate increasing QoS demands from end-users (QoS failures), depending on spatial ortemporal locality. For example, links around aconference room may have to relay too muchdata/video traffic during the session. Likewise,relay links outside the room may fail to support allattendees’ voice-over-IP calls during a sessionbreak. By reassociating their radios/channels withunderutilized radios/channels available nearby,links can avoid communication failures[6].iii) Coping with heterogeneous channelavailability:Links in some areas may not be able toaccess wireless channels during a certain timeperiod (spectrum failures) due to spectrum etiquetteor regulation. For example, some links in a WMNneed to vacate current channels if channels arebeing used for emergency response near thewireless links (e.g., hospital, public safety). Suchlinks can seek and identify alternative channelsavailable in the same area.Motivated by these three and otherpossible benefits of using reconfigurable mr-WMNs, in the remainder of this paper, we wouldlike to develop a system that allows mr-WMNs toautonomously change channel and radioassignments (i.e., self-reconfigurable) to recoverfrom the channel-related link failures.3.4 QoS-satisfiability EvaluationAmong a set of feasible plans, ANRS nowneeds to identify QoS-satisfying reconfigurationplans by checking if the QoS constraints are met
  5. 5. B.Sindhu, V.Bharathi, N.Karthikeyan / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.092-09996 | P a g eunder each plan. Although each feasible planensures that a faulty link(s) will use non faultychannels and maintain its connectivity, some plansmight not satisfy the QoS constraints or may evencause cascaded QoS failures on neighboring links.To filter out such plans, ANRS has to solve thefollowing challenges.1) Per-link bandwidth estimation:For each feasible plan, ANRS has to checkwhether each link’s configuration change satisfiesits bandwidth requirement, so it must estimate linkbandwidth. To estimate link bandwidth, ARSaccurately measures each link’s capacity and itsavailable channel airtime. In multihop wirelessnetworks[8] equipped with a CSMA-like MAC,each link’s achievable bandwidth (or throughput)can be affected by both link capacity and activitiesof other links that share the channel airtime.Even though numerous bandwidth-estimation techniques have been proposed, theyfocus on the average bandwidth of each node in anetwork or the end-to-end throughput[7] of flowswhich cannot be used to calculate the impact ofper-link configuration changes. By contrast, ANRSestimates an individual link’s capacity based onmeasured (or cached) link-quality information—packet-delivery ratio and data-transmission ratemeasured by passively monitoring thetransmissions of data or probing packets.Here, we assume that ANRS is assumed tocache link-quality information for other channelsand use the cached information to generatereconfiguration plans. If the information becomesobsolete, ARS detects link failures and triggersanother reconfiguration to find QoS-satisfiableplans.2) Examining per-link bandwidthsatisfiability :Given measured bandwidth and bandwidthrequirements, ANRS has to check if the new linkchange(s) satisfies QoS requirements. ARS definesthe expected busy airtime ratio of each link tocheck the link’s QoS satisfiability.3) Avoiding cascaded link failures:Besides the link change, ANRS needs tocheck whether neighboring links are affected bylocal changes (i.e., cascaded link failures). Toidentify such adverse effect from a plan, ANRSalso estimates the QoS satisfiability of links onehop away from member nodes whose links’capacity can be affected by the plan. If these one-hop-away links still meet the QoS requirement, theeffects of the changes do not propagate thanks tospatial reuse of channels. Otherwise, the effects oflocal changes will propagate, causing cascadedQoS failures.3.5 Choosing the Best PlanANRS now has a set of reconfigurationplans that are QoS-satisfiable and needs to choose aplan within the set for a local network to haveevenly distributed link capacity. However, toincorporate the notion of fair share into theplanning, ANRS needs to address the followingchallenges.i. Quantifying the fairness of a planii. Breaking a tie among multiple plansi) Quantifying the fairness of a plan:ANRS has to quantify the potentialchanges in link-capacity distribution from a plan.To this end, ANRS defines and uses a benefitfunction that quantifies the improvement of channelutilization that the reconfiguration plan makes.Specifically, the benefit function is defined as , therelative improvement in the airtime usage of radio ,and the number of radios whose has changed fromthe plan. This definition allows the benefit functionto quantify the overall change in airtime usage,resulting from the reconfiguration plan.ii) Breaking a tie among multiple plans:Multiple reconfiguration plans can havethe same benefit, and ARS needs to break a tieamong them. ANRS uses the number of linkchanges that each plan requires to break a tie.Although link configuration changes incur a smallamount of flow disruption (e.g., in the order of 10ms) the less changes in link configuration, the lessnetwork disruption.ANRS favors a plan that reconfigureslinks to have 50% available channel airtime. If aplan reconfigures a WMN to make the links heavilyutilized while idling others (e.g., plan 2), then thebenefit function considers the plan ineffective,placing the plan in a lowly ranked position.4. Performance EvaluationANRS effectively reconfigures thenetwork on detection of a failure, achieving morebandwidth than static assignment and localrerouting. ANRS accurately detects a link’s QoS-failure using link-quality monitoring informationand completes network reconfiguration. ANRS alsoimproves channel efficiency. ANRS identifiesfeasible changes that help avoid a local link failurebut maintain existing network connectivity as muchas possible. ANRS is implemented as an agent inboth the MAC layer and a routing protocol. Itperiodically collects channel information fromMAC and requests channel switching or link-association changes based on its decision. At thesame time, it informs the routing protocol ofnetwork failures or a routing table update.
  6. 6. B.Sindhu, V.Bharathi, N.Karthikeyan / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.092-09997 | P a g eThe simulation topology is set up toanalyze the performance of ANRS. It is alsoexpected that the ANRS effectively reconfiguresthe network around a faulty link improving bothnetwork throughput and channel efficiency. Bycontrast local rerouting causes degradation in thechannel efficiency due to the use of a detour path.The performance parameters that can be obtainedthrough the NS2 Trace Analyser are as follows Throughput Channel efficiency DelayThroughput: Throughput is the rate at which anetwork sends and receives data. Throughput israted in terms bits per second (bit/s). Throughput,Tp=Pa/Pf, Pa = packets received, Pf is the amountof forwarded over certain time interval.Channel Efficiency: channel efficiency is a rate atwhich a network delivers maximum informationfrom source to destination.Delay: Delay refers to the time taken for a packetto be transmitted across a network from source todestination. Delay, D = Td - Ts Td = packetreceived time at destination Ts = packet time atsource.4.1 Experimental ResultsWe evaluated the improvements achieved byANRS, including throughput and channelefficiency, QoS satisfiability, and reduction ofripple effects.A) Throughput and Channel-Efficiency Gains:We first study throughput and channel-efficiency gains via ANRS’s real-timereconfiguration. We run one UDP flow at amaximum rate over a randomly chosen link in test-bed while increasing the level of interference every10s.Fig 1: ThroughputWe also set the QoS requirement of every link to 6Mb/s and measure the flow’s throughputprogression every 10 s during a 400-s run.. Notethat we do not intentionally run a greedy algorithmin this single-hop scenario because its effect issubsumed by ANRS.ANRS effectively reconfigures thenetwork on detection of a failure, achieving 450%and 25.6% more bandwidth than static-assignmentand local re-routing, respectively. ANRS accuratelydetects a link’s QoS-failure using link-qualitymonitoring information, and completes networkreconfiguration (i.e., channel switching) within 15seconds on average, while the static-assignmentexperiences severe throughput degradation. Notethat the 15- second delay is due mainly to link-quality information update and communicationdelay with a gateway, and the delay can beadjusted. Further, within the delay, actual channelswitch delay is less than 3 ms, which causesnegligible flow disruption. On the other hand, thelocal re-routing improves the throughput by using adetour path, but still suffers from throughputdegradation because of an increased loss rate alongits detour path.Fig 2: Improvement in throughputANRS also improves channel efficiency(i.e., the ratio of the number of successfullydelivered data packets to the number of total MACframe transmissions) by more than 90% over theother recovery methods. Using the data collectedduring the previous experiment, we derive channelefficiency of the UDP flow by counting the numberof total MAC frame transmissions and the numberof successful transmissions.ANRS reconfigures a wireless meshnetwork to meet different QoS requirements.Before each reconfiguration, the gray areas canonly accept 1 to 9 UDP flows. On the other hand,after reconfiguration, the network in the areas canadmit 4 to 15 additional flows, improving theaverage network capacity of the gray areas by 3.5times.. Further, within the delay, actual channelswitch delay is less than 3 ms, which causesnegligible flow disruption. On the other hand, thelocal re-routing improves the throughput by using adetour path, but still suffers from throughput
  7. 7. B.Sindhu, V.Bharathi, N.Karthikeyan / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.092-09998 | P a g edegradation because of an increased loss rate alongits detour path.ANRS improves channel efficiency by upto 91.5% over the local rerouting scheme thanks toits online channel reconfiguration. On the otherhand, using static channel assignment suffers poorchannel utilization due to frame retransmissions onthe faulty channel. Similarly, the local reroutingoften makes traffic routed over longer or low link-quality paths, thus consuming more channelresources than ANRS.Fig 3: Channel-efficiency gainsEffectiveness of QoS-aware planning: Wemeasured the effectiveness of ARS in meeting thevarying QoS requirements in a mr-WMN. Weinitially assign symmetric link capacity as shown inthe channel assignment of the grid topology. Then,while changing the QoS constraints in gray areas atdifferent times (i.e., T1 to T5), we evaluate theimprovement of available capacity that ANRS cangenerate via reconfiguration.Impact of the benefit function: We alsostudied the impact of the benefit function on theARS’s planning algorithm. We conducted the sameexperiment as the previous one with differentvalues of δ in the benefit function. ANRS allows tokeep local channel-efficiency high. By contrast, alow value (0.4) can deliver more availablebandwidth (on average, 1.2 Mbps) than when thehigh value is used, since ANRS tries to reservemore capacity.Impact of the reconfiguration range: Weevaluated the impact of the reconfiguration range.We used the same experiment settings as theprevious one and focused on reconfigurationrequests at T(k). As we increase the hop count (k)from a faulty link(s), we measure the capacityimprovement achieved by the reconfigurationplans. In addition, we calculate the capacity gainper change as the cost-effectiveness ofreconfiguration planning with different k values.Fig. 4 plots the available capacity of the faulty areaafter reconfigurations. As shown in the figure,ANRS can improve the available links’ capacity byincreasing the reconfiguration range. However, itsimprovement becomes marginal as the rangeincreases. This saturation results mainly from thefixed number of radios of each node.In other words, the improvement isessentially bounded by the total capacity ofphysical radios. Furthermore, becausereconfiguration plans with a larger range arerequired to incur more changes in network settings,the bandwidth gain per change significantlydegrades (e.g., capacity gain per change at the hop-count of 4 in Fig. 4). We also observed the similarresults in other reconfiguration requests(T2,T3,T4), but omitted them for brevity.Fig 4: Capacity of nodes in ARS methodB) QoS Satisfaction Gain:ANRS enhances the chance to meet thevarying QoS demands. We use static WCETTrouting metric that finds a path with diversechannels and ARS for reconfiguration.QoS-aware reconfiguration planning inANRS improves the chance for a WMN to meet thevarying QoS demands, on average, by 200%. ARSeffectively discovers and use idle channel throughreconfiguration, thus satisfying QoS demands by upto three times more than static-assignmentalgorithm.C) Avoidance of Ripple Effects:We also studied ANRS’s effectiveness inavoiding the ripple effects of networkreconfiguration. Here three failure recoverymethods have been used (i.e., local rerouting,greedy, and ANRS). Since ANRS considers theeffects of local changes on neighboring nodes via, iteffectively identifies reconfiguration plans thatavoid the ripple effects. Under ANRS, the networkperforms reconfiguration and recovers an average98% of the degraded throughput (4.8 Mb/s).
  8. 8. B.Sindhu, V.Bharathi, N.Karthikeyan / International Journal of Engineering Research andApplications (IJERA) ISSN: 2248-9622 www.ijera.comVol. 3, Issue 3, May-Jun 2013, pp.092-09999 | P a g e5. CONCLUSIONThis paper presented an autonomousnetwork reconfiguration system (ANRS) thatenables a multiradio WMN to autonomouslyrecover from wireless link failures. ANRSgenerates an effective reconfiguration plan thatrequires only local network configuration changesby exploiting channel, radio, and path diversity.Furthermore, ANRS effectively identifiesreconfiguration plans that satisfy applications’ QoSconstraints, admitting up to two times more flowsthan static assignment, through QoS awareplanning. In multihop wireless networks equippedwith a CSMA-like MAC, each link’s achievablebandwidth (or throughput) can be affected by bothlink capacity and activities of other links. ANRSidentifies feasible changes that help avoid a locallink failure but maintain existing networkconnectivity as much as possible. It can improvethe available links’ capacity by increasing thereconfiguration range. ANRS’s localreconfiguration improves network throughput andchannel efficiency. Our experimental evaluationson a Linux-based implementation and ns2-basedsimulation have demonstrated the effectiveness ofANRS in recovering from local link-failures and insatisfying applications’ diverse QoS demands.References[1] I. Akyildiz, X. Wang, and W. Wang,―Wireless mesh networks: A survey,‖Comput. Netw., vol. 47, no. 4, pp. 445–487, Mar. 2005.[2] P. Kyasanur and N. Vaidya, ―Capacity ofmulti-channel wireless networks: Impactof number of channels and interfaces,‖ inProc. ACM MobiCom, Cologne,Germany, Aug. 2005.[3] K. Ramanchandran, E. Belding-Royer,and M. Buddhikot, ―Interference- awarechannel assignment in multi-radio wirelessmesh networks,‖ in Proc. IEEEINFOCOM, Barcelona, Spain, Apr. 2006,pp. 1–12.[4] R. Draves, J. Padhye, and B. Zill,―Routing in multi-radio, multi-hopwireless mesh networks,‖ in Proc. ACMMobiCom, Philadelphia, PA, Sep. 2004,pp. 114–128.[5] D. Aguayo et al., ―Link-levelmeasurements from an 802.11b meshnetwork,‖ in Proc. ACM SIGCOMM2004, Aug. 2004.[6] A. Willsky, ―A Survey of Design Methodsfor Failure Detection in DynamicSystems,‖ Automatica, vol. 12, pp. 601–611, 1976.[7] A. Brzezinski, G. Zussman, and E.Modiano, ―Enabling distributedthroughput maximization in wireless meshnetworks: A partitioning approach,‖ inProc. ACM MobiCom, Los Angeles, CA,Sep. 2006, pp26–37.[8] D. S. D. Couto, D. Aguayo, J. Bicket, andR. Morris, ―A high-throughput path metricfor multi-hop wireless routing,‖ in Proc.ACM MobiCom, San Diego, CA, Sep.2003, pp. 134–146.