A Study of Routing Techniques in Intermittently Connected MANETs


Published on

A Mobile Ad hoc Network (MANET) is a self-configuring infrastructure less network of mobile devices connected by wireless. These are a kind of wireless Ad hoc Networks that usually has a routable networking environment on top of a Link Layer Ad hoc Network. The routing approach in MANET includes mainly three categories viz., Reactive Protocols, Proactive Protocols and Hybrid Protocols. These traditional routing schemes are not pertinent to the so called Intermittently Connected Mobile Ad hoc Network (ICMANET). ICMANET is a form of Delay Tolerant Network, where there never exists a complete end – to – end path between two nodes wishing to communicate. The intermittent connectivity araise when network is sparse or highly mobile. Routing in such a spasmodic environment is arduous. In this paper, we put forward the indication of prevailing routing approaches for ICMANET with their benefits and detriments

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

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide

A Study of Routing Techniques in Intermittently Connected MANETs

  1. 1. International Journal of Research in Computer ScienceeISSN 2249-8265 Volume 3 Issue 3 (2013) pp. 35-43www.ijorcs.org, A Unit of White Globe Publicationsdoi: 10.7815/ijorcs. 33.2013.066www.ijorcs.orgA STUDY OF ROUTING TECHNIQUES ININTERMITTENTLY CONNECTED MANETSS. Ramesh1, P. Ganesh Kumar21Faculty/CSE, Anna University, Regional Centre Madurai, INDIAEmail: itz_ramesh87@yahoo.com2Professor/ECE, KLN College of Engineering, Sivagangai, INDIAEmail: ganesh_me@yahoo.comAbstract: A Mobile Ad hoc Network (MANET) is aself-configuring infrastructure less network of mobiledevices connected by wireless. These are a kind ofwireless Ad hoc Networks that usually has a routablenetworking environment on top of a Link Layer Ad hocNetwork. The routing approach in MANET includesmainly three categories viz., Reactive Protocols,Proactive Protocols and Hybrid Protocols. Thesetraditional routing schemes are not pertinent to the socalled Intermittently Connected Mobile Ad hocNetwork (ICMANET). ICMANET is a form of DelayTolerant Network, where there never exists a completeend – to – end path between two nodes wishing tocommunicate. The intermittent connectivity araisewhen network is sparse or highly mobile. Routing insuch a spasmodic environment is arduous. In thispaper, we put forward the indication of prevailingrouting approaches for ICMANET with their benefitsand detriments.Keywords: MANET, ICMANET, Delay TolerantNetworks, Routing Schemes, Performance Parameters.I. INTRODUCTIONMANET is well – thought – out to be the rapiddeployment of independent mobile users. Substantialexamples include establishing survivable, efficient anddynamic communication for emergency/rescueoperations, disaster relief efforts, and militarynetworks. Such network setups cannot count oncentralized and organized connectivity and can beconceived as applications of Mobile Ad hoc Networks.A MANET is an autonomous collection of mobileusers that communicate over relatively Band Widthconstrained wireless links. Since the nodes are mobile,the network topology may change rapidly andunpredictably over time. In quintessence, the networkis decentralized; where all network activity includingdiscovering the topology and delivering messagesmust be executed by nodes themselves, (i.e.) routingfunctionality will be assimilated into mobile nodes.The routing methodologies in such network are madethrough traditional routing protocols like AODV (Adhoc on demand Distance Vector), DSR (DestinationSource Routing) etc. Ample routing algorithms havebeen proposed for MANET. The MANET paves to anew form of network called the ICMANET.ICMANET is painstaking to be one of the newareas in the field of wireless communication. Networksunder this class are potentially deployed in challengedenvironments using isolated mobile devices withlimited resources. These are emerging as a promisingtechnology in applications such as in WildlifeManagement, Military Surveillance, UnderwaterNetworks and Vehicular Networks. ICMANET, alsoknown as the Delay Tolerant Network (DTN), istypically different from traditional Mobile Ad hocNetworks (MANETs), which means that in the latter;communication between two nodes is possible at anytime via a path of intermediate nodes although thispath may vary with time. However, in an ICMANETpaths between to nodes have to be established only bymultihop paths that span over space and time. In otherwords, there is no end – to – end path between the twoat any given instant. In this paper, we deliver a studyof possible routing schemes in ICMANET. Thetraditional routing scheme that forms a basis for otherrouting schemes in ICMANET is the Flooding basedrouting. In this, one node sends packet to all othernodes in the network. Each node acts as both atransmitter and a receiver. Each node tries to forwardevery message to every one of its neighbors [15]. Theresults in every message eventually delivered to allreachable parts of the network. In rest of this paper, wewill describe the various routing protocols availablefor ICMANET.II. EPIDEMIC ROUTINGThe Epidemic Routing protocol is a Flooding basedrouting protocol which states that periodic pair – wiseconnectivity is necessitated for message delivery. Theprotocol banks on immediate dissemination ofmessages across the network. Routing occurs based onthe node mobility of carriers that are within distinctiveposition of the network as in Figure 1 and Figure 2.The protocol is designed in such a way that each host
  2. 2. 36 S. Ramesh, P. Ganesh Kumarwww.ijorcs.orgmaintains a buffer. The buffer holds messages whoseorigin is in that particular host, as well as messages ofthe secondary hosts. It also sustains a hash table. Thehost value catalogs the directory of messages that arekeyed by inimitable identifier associated with eachmessage.A Summary Vector [13] is associated with the host,to signpost which entries in table are set. The routingcomprises an anti – entropy session. When two hostsare within the same communication range, one withthe least identifier inducts the anti – entropy session tothe one with larger identifier. Under this session twohosts barter their summary vector, to explore themessages that were not viewed by each other. Laterthey request copies of messages that were notreconnoitered by them. The routing scheme includes amessage identifier, hop count and anacknowledgement request. (i) Message identifier –unique 32 – bit number; is a combination of host IDand a locally generated message ID (i.e.) 16 bit. (ii)Hop count – value implies the distribution ofmessages; similar to Time to Live (TTL) field. (iii)acknowledgement request – acts a signal to thedestination requesting to endow acknowledgement formessage delivery.It is studied [13] that goals of this routing schemeinclude:i. Maximum message delivery rate.ii. Minimum message latency.iii. Minimal resources consumption.III. BEACONLESS ROUTINGThe beaconless routing protocol is grounded on thehypothesis where there never exists an intervallicdiffusion of beacons into the network. The conventionrelies upon the acquaintance of geographical positionthat aids in the reduction of overhead. Routingprimarily makes a choice of forwarding node in adispersed modus amidst its neighbors, without anyform of erudition about their location prevalence. Theforwarding of messages is accomplished by smearingDynamic Forwarding Delay (DFD) that indicts one ofthe possible relay nodes, relays the message packets.Relay nodes refers to the intermediate nodes in anetwork.The algorithm involves the following sequence ofsteps in routing:i. As a node broadcasts a packet, its neighboring nodereceives it.ii. DFD claims that only one of relay nodes transmitsthe packet.iii.Nodes with least delay deliver the packet first andsecondary relay nodes perceive the delivery andabandon its reserved transmission of the samepacket.From [12], it is known that some of the assumptionsare allied to BLR. (i) Nodes have a predeterminedknowledge of its own position via GPS, Galileo or anyother kind of positioning services. (ii) Nodes are awareof two parameters viz., maximum delay and maximumtransmission.From [12], it is eminent that, if a sender holds apacket, it ascertain the destination, stores itstopographical coordinates, associated with its ownposition, in the header of the packet. Each and everyrelay node supersedes previous node’s position by itscurrent position in packet header before forwarding.Packet is advertised to all neighbouring nodes. Onreception of packet, the only offered information of theintermediate node is its own position, position ofprevious and destination node that were extracted frompacket header. Nodes within forwarding area applyDFD, prior to broadcasting of packets and those nodesthat are outside this area discard the received packet.Destination nodes use passive acknowledgement. Theprevious transmitting node also finds that the packetsare further delayed and clinches that other node hasreceived it successfully. The algorithm persists untildestination is reached and destination has to sendacknowledgement on reception. BLR applies anadaptive and restricted reactive protocol based onAODV in the vicinity of destination primarily to dealwith inaccuracies.Figure 1: Routing of Messages at t==0SABCDAt t==0
  3. 3. A Study of Routing Techniques in Intermittently Connected MANETs 37www.ijorcs.orgFigure 2: Routing of Messages at t > 0The delay function is calculated using the formula[12],Add_delay = Max_delay ((r-p)/r)Where r = maximum transmission radius.P = progress.Significance of BLR is to minimize delay and mayoperate at promiscuous node. The performance of BLRis based on the greedy mode which insists thebroadcast of all packets and do not hold any recoverystrategy when no node is found to be in the forwardingarea and the packets are just dropped. The advantageof BLR is found to be, extensive battery power usage.IV. CONTEXT AWARE ROUTINGThe Context Aware Routing (CAR) algorithmpaves the forethought of asynchronous communicationin ICMANET. The algorithm endows a basement oforganizing the messages in the network. It addressesthat the nodes are able to exploit the contextinformation to make local decisions which imparts thegood delivery ratios and latencies with less overhead.CAR is pain staked as a general framework topredict and evaluate context information for superiordelivery of messages. The context here refers to theaspects of system. The delivery process, in thisalgorithm is based on the presence of receiver withinthe identical set of nodes. The prospect of delivery ofmessages is generated from the context information.The algorithm states that if both source and destinationnodes are in the same network, message is deliveredsynchronously. A proactive routing is used and ispresumed that every single node host transmits jointlythe information related to synchronously routing and alist of delivery probability for the other host nodes. Inthe mode of asynchronously routing, each host holds alist of entries whose tuples are destination, best host,and delivery probability. As a host is selected to be acarrier and receives the message, it inserts into thebuffer.The prediction of context information is as follows:[11]i. The host calculates its own delivery probabilitybased on (a) Prediction of future values. (b)Deriving estimated values with utility theory.ii. Host maintains a forwarding table with tuples nexthop and delivery probability for all knowndestinations.iii.Each and every host uses the prediction of deliveryprobability at times of temporary disconnection.iv. When a host does not hold any information aboutthe receiver, messages are transmitted to nodes withhigher mobility.v. When carriers meets node with higher deliveryprobability during its mobility, message is deliveredto it.The evaluation of context information involves:i. Local evaluation: - defines a static hierarchy or usesa pre-emptive methodology.ii. Significance based evaluation: - uses a utilityfunction.Maximise {f (U (xi)) = ∑ 𝑤𝑖𝑛𝑖=1 Ui (xi)} [11]The performance of CAR is depicted as the numberof messages exchanged is more or less constantregardless of buffer size, demonstrating its scalability.V. BROWNIAN GOSSIPThe Brownian gossip is an amalgamation of gossipand the random node mobility which provides ascalable geographical routing. In this routing, eachnode forwards the query related to other nodesinformation with certain values of probability.Gossiping is a resourceful approach for informationdissemination and is done with a probability viz.,Pgossip. The probability value makes certain that thequery can reach the secondary nodes in the networkwith highest probability. The algorithm is depictedFigure 3.The reply message is routed to the query originator.To make a correct decision of propagating direction ofSBACDAt t >0
  4. 4. 38 S. Ramesh, P. Ganesh Kumarwww.ijorcs.orga query, hints about the location of destination is used.The hints are obtained from the carriers. During thenode mobility, several nodes will be encountered andthis memory of encounters acts as hints.Figure 3: Working of BrownianAssume that a node A receives a query, it hasencountered node D at location (m, n) at time tD unitsin the past. Node A forwards query to location (x, y),as the nodes A, B meet, they acquire information aboutthe geographical position and updates its local cachesas follows:Node A updates as,{Node B, ti (XB, YB), Is Neighbor TRUE} [10]Similarly node B updates its cache. The updating isperformed periodically based on the frequency ofbeacon signals. When A, B are not immediateneighbours’, the neighbour field is set FALSE. Whenthese nodes encounter new nodes, they gossip theirmemory of encounters with probability Pgossip, [10]which is chosen on the ground of (i) mobility patternof nodes (ii) traffic pattern in network .The querypacket contains of{Source S, Locations (Xs, Ys), Target D, tm, Hopsh}[10]When the propagated query reaches the receiver, itreplies back its current location (i.e.) (XD, YD) to thesender.The value of k copies is chosen as follows:K = (min(𝑡𝑐𝑢𝑟𝑟𝑒𝑛𝑡 − 𝑡𝑙𝑎𝑠𝑡−𝑒𝑛𝑐𝑜𝑢𝑛𝑡𝑒𝑟)𝑐, kmax) [10]Where c, kmax = constants.The Pgossip is determined by using,Pigossip= max {f (Query Counti), g (AvgSpeedi)} [10]Where Query Counti = number of times queriedwithin a time interval in past.AvgSpeedi = average speed with whicheach node has travelled inpast.The gossip should take place at a time intervalof2RV, where R = communication range, V= Speed ofnode that is ready to gossip. The main advantage ofgossip is that the overhead is lowered as the networkdoes not transmit all the queries that were received.VI. MOBILITY PROFILE BASED ROUTINGThe mobility profile based routing addresses, a hub– level routing method and two versions of user – levelrouting methods. The routing involves a SOLAR–HUB (Sociological Orbit aware LocationApproximation and Routing) which manipulates theuser profiles that aids in hub – level routing. Themultipath version of SOLAR–HUB delivers messageto most visited and second most visited destinationhubs via k – neighbor. The two user level routingprotocols are viz., Static SOLAR–KSP and DynamicSOLAR–KSP grounded on contact probability and k–shortest paths and KSP is K- shortest paths.The SOLAR–HUB algorithm [14] states that sourcereaches destination by forwarding data to k of its ownneighbors. Every user computes delivery probability toevery hub as,Delivery probability = max (travel probability, max(contact probability (hub to k visit)*travelprobability)) [9]The Static SOLAR–KSP algorithm computes thecontact probability P (U, V) and constructs a weightedgraph W (U, V).W (U, V) =𝑙𝑜𝑔(1𝑃(𝑈,𝑉)) [9]The KSP is determined by [9] (i) initially path withminimum total weight is chosen. (ii) Each path hasdifferent next hop, then the KSP are given deliveryprobability. The Dynamic SOLAR–KSP algorithm is acombination of Static hub based information andinvolves dynamic selection of next hop.Regarding performance the HUB protocols areeminent as follows:i. SOLAR–HUB shows maximum throughput andhigher overhead.ii. As user density increases, D–SOLAR KSP and S–SOLAR KSP shows increase in delivery probabilityand decrease in end – to – end delay.iii.As number of hub increases and user densitydecreases, D–SOLAR KSP and S–SOLAR KSPshows a decrease in data throughput and increase inend – to – end delay.iv. The overhead of D–SOLAR KSP is minimal incontrast to S–SOLAR KSP.VII. DIRECTION BASED GEOGRAPHICROUTINGThe Direction Based Geographic Routing (DIG)algorithm is grounded on geographic location ofpackets that are routed in an average approximate idealpath towards destination. The next hop in DIG ischosen by considering the following factors [8]:
  5. 5. A Study of Routing Techniques in Intermittently Connected MANETs 39www.ijorcs.orgi. Distance between relay node and destination node.ii. Moving direction of the mobile node.The algorithm postulates that when two nodesencounter each other, the nodes exchange theknowledge of their current location, moving directionand the packets. The packets are forwarded to nodeswhose distance and moving direction are closest todestination. The algorithm includes three steps [8]:Source forwards each part of packet to a number ofneighbour nodes when it meets.i. d>T (T is threshold distance) node chooses movingdirection α that is between [θ-ξ, θ+ξ], ξ=Tn.arcsin�rd� ≤ π2� . [8]ii. d<T node A forwards packet to node B if αB<αA.[8]DIG has decreased delay as the resources requiredare less.VIII. SINGLE COPY ROUTINGThe single copy routing, from its nomenclature itpostulates that only a single copy of message packet iscarried to destination. This routing (i) employed whenlow resource usage is critical (ii) improves the designof routing schemes that uses multiple copies.The routing schemes of single copy are, [7]i. Direct transmission in which packets are delivereddirectly to destination.ii. Randomized routing in which one custodianforwards to another one it meets with certainprobability P ∈(0, 1].iii.Utility based routing that ground upon theusefulness of a node in delivering message toanother node.iv. Seek and focus (hybrid) has three phases [7]a. Seek phase: - based on Utility<Uf (focusthreshold) and performs randomized routing.b. Focus phase: - based on Utility>Uf, receives themessage then resets a timer and later performsutility based function.c. Reseek phase: - if tfocus expires, the timer ismodified and randomized routing is performedbased on utility functions.v. Oracle – based optimal algorithm is aware ofmovement with which, an optimal set of forwardingdecisions are made whose delivery probabilityoccurs at minimum time.IX. MULTIPLE COPY ROUTINGThe multiple copy scheme deals with themechanism of spraying a few copies of message andthen routing each copy in isolated manner to thedestination. The advantage of this scheme is stated as[6]i. Less transmission.ii. Less delays.iii.Higher scalability.The design goals of multiple copy case is to, [6]i. Perform fewer transmissions at all circumstances.ii. Quicker message delivery.iii.Maximum amount of messages generated are to bedelivered.iv. Achieve high scalability.v. Providing simple and minimum knowledge aboutnetwork.The algorithm that holds multiple copy case routingare Spray & Wait and Spray & Focus. The Spray andWait involves spraying multiple copies into thenetwork and waiting for the destination to be reachedbefore delivering the packet. The Spray and focusengross spraying of fewer copies of message packetsinto network and forwards packet based on the utilityfunction if and only if utility function of node B toreach destination is greater than the sum of the utilityfunction of node A to reach destination and the utilitythreshold.X. SEMI PROBABLISTIC ROUTINGSemi Probabilistic Routing (SPR) algorithmconsiders that the network is partitioned into tinyportions that have a stable topology. The protocolupholds the information about host mobility andconnectivity changes for more accurate messageforwarding. Information is also used to handle thebuffer space. In this, the mobile node maintains h hopsfrom itself. When the receiving node is present withinthe sane zone of sending node, datas are deliveredsynchronously whereas when synchronous deliveriesof messages are not admitted, datas are forwardedtowards a group of nodes that has the highest deliveryprobability. The routing is a combination of bothprobabilistic routing with deterministic decisions. Thealgorithm has profile information that is described as aset of attributes exploiting the overall aspects of thesystem.The probability of message delivery is calculatedfor node i for data j as follows: [5]Pij= �1 , 𝑖𝑓 𝑛𝑜𝑑𝑒𝑖 𝑖𝑠 𝑎 𝑟𝑒𝑐𝑒𝑖𝑣𝑒𝑟 𝑜𝑓 𝑑𝑎𝑡𝑎 𝑗𝑃𝑖𝑗𝑑 , 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒[5]The Pijd denotes the intermediate node and iscalculated as,Pijd= wcdc Ucdc, i + wcol, Ucol, ij [5]
  6. 6. 40 S. Ramesh, P. Ganesh Kumarwww.ijorcs.orgWhere Ucdc, i = change degree of connectivity.Ucol, ij = history of co-location of I with dataj.Ucdc, i =|𝑛(𝑡−𝑇)∪ 𝑛(𝑡)|− |𝑛−(𝑡−𝑇)𝑛 𝑛(𝑡)||𝑛(𝑡−𝑇)∪𝑛(𝑡)[5]Ucol, ij = �1, 𝑖𝑓 𝑗′𝑠𝑟𝑒𝑐𝑒𝑖𝑣𝑒𝑟 𝑖𝑠 𝑖𝑛 𝑖′𝑠 𝑧𝑜𝑛𝑒0, 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒[5]The data j is sent to nodes in the routing zone ofnode i when the delivery probability is greater than theforwarding threshold. This algorithm shows maximummessage delivery and minimum blind messageforwarding.XI. CONTENTION BASED ROUTINGThe Contention Based Routing postulates that theefficiency of routing can be achieved only by takinginto account the contention and dead end [4]. TheSpray Select and Focus provides a better performanceconsidering the contention and dead ends. Thecontention is referred to be as the competition forresources. It is a condition, where two or more nodesattempts to forward messages across networksimultaneously. The operation of Spray Select Focus isdepicted pictorially in Figure 4. Source node sendsmultiple copies of message to various relay nodes.When one of the relay nodes reaches destination, theother copies are dropped [4]. Hence the route isminimized and contention is avoided.The algorithm holds three phases: [4]i. In Spray phase L message copies are sprayed bysource.ii. Select phase selects node and finds a shortest routewith the hop distance to destination.iii. In Focus phase, the utility of node X to thedestination Y is calculated.Figure: 4 Spray Select FocusNode A forwards to node B only if the utilityfunction of B to D is greater than A (i.e) A broadcaststo relay node B if and only if UB(D)>UA(D) [4].Thedead end is a situation, that occurs due to hardware orpower failure and hence the nod gets struck. Thisalgorithm proposes two ways [4] to recover from deadends.i. By routing copies using bypass recovery.ii. No routing is done; the focus phase will transmitthe copies directly.XII. SPRAY AND HOPThe Spray and Hop is a routing protocol that holdstwo phases namely, Spray phase that sprays few copiesof message into the network. Hop phase which occursafter the spraying phase, a node that was not able tofind the destination, switches to the hop phase.Spray Phase:When a message is generated at the source node, itcreates k forwarding tokens. The spray phasemaintains a summary vector and its fields are shown inFigure 7 [3].Spray phase includes certain rules for forwardingthe messages and is described as follows:i. When two nodes encounter each other, theyexchange their summary vectors and check formessages that are in common.ii. A node that has a message copy has k forwardingtokens. When that encounters a node withoutmessage, forwards the message copy and handovercertain forwarding tokens.iii.A node transmits data to another node that hasgreater remaining power.iv. When more than one node possesses the samemessage, node with greater residual energy has theresponsibility to forward message.v. When a node has data packet and has a singletoken, it forwards according to the hop phase.Spraying is of two forms viz.,a. Source Spray, in which the source node issues oneforward token each time and retains k-1 tokens foritself.b. Binary Spray, in which the source node issues halfof its token at each forward.Hop Phase:The hop phase occurs only when one forwarding tokenis left with the source node. The utility function is usedto make a choice for relay node based on the residualpower and rediscovery interval. This techniqueoutperforms when the number of transmissions areconsidered. It is said that as it uses utility function [3]to choose the relay node rather than waiting toencounter the destination, it is estimated that it hasbetter performance than spray and wait.SAcccAcAcAcB DreachIgnored as oneof the copiesreaches thedestination
  7. 7. A Study of Routing Techniques in Intermittently Connected MANETs 41www.ijorcs.orgXIII. SPRAY AND WAITThe Spray and Wait is a scheme that sprays into thenetwork a fewer number of message copies and waitsuntil one of these nodes that holds the copies reachesthe destination. It is simple to implement and can beoptimized to achieve the depicted performance. TheSpray and wait has certain features as (i) Performingfewer transmissions, (ii) Generating low contention(iii) Achieving better delivery delay (iv) Highscalability.The protocol has two phases [2]a. Spray phase: L message copies are initially spreadby the source and other L distinct relay nodesreceives the copies.b. Wait phase: When destination is not determinedduring the spray phase, the relay node wait untilthey determine the destination and then performsdirect transmission.In this, when enough copies are spread into thenetwork, it is guaranteed that at least one node willreach the destination. At this point, the source nodestops spraying and lets each custodian node to performthe direct transmission. In choosing the L copies to besprayed initially two ways were used. The simplestway is to spread the L copies to all L distinct nodesthat source node encounters initially. The secondarytechnique is to use binary spray and wait [2], in whichthe source node handoffs n2� copies to the relay nodesit encounters by keeping n2� copies for itself. Whenthe source node is left with only one copy of messageit performs direct transmission. The spray and wait [2]technique, that is highly robust and scalable. It is statedthat it is the only routing scheme that achieves bothlow delivery delay and higher energy efficiency. It alsoholds fewer transmissions for a single message to bedelivered.XIV. LOCATION AWARE ROUTING ON DELAYTOLERANT NETWORKS – LOCATIONDISSEMINATIONThe Location Aware Routing On Delay tolerantnetworks – Location Dissemination (LAROD-LODIS)is a new form of routing that complements efficientrouting in the Intermittently Connected Mobile Ad hocNetworks. The LAROD-LODIS, as the name indicatesit is an amalgamation of LAROD-Location AwareRouting for Delay Tolerant Networks and LODIS-Location Dissemination Services. The amalgamationof these techniques suits ICMANET in a well – versedway.LAROD [1] routes packet only with partialknowledge of the geographic positions. To impose lowoverhead, it uses beaconless protocol to route packets.It combines the routing scheme with store-carry-forward technique and uses greedy packet forwardingwhenever required.To broadcast a message, the custodian transmitsmessage. All nodes that stay within the specifiedtransmission range receive the packet. A delay timer isset for each custodian; node whose delay timer expiresfirst becomes the new custodian. New custodianforwards message. Relay nodes hearing this, assumesthat new custodian has taken form. If no suchtransmission is heard, the old custodian retransmitswith some retransmission time tr. It is stated that, allnodes do not hear the broadcast; hence duplication ofpackets may occur. When paths of two copies crosseach other, only one forwards, while the other isdiscarded. Each packet has a TTL [1], when it expiresthe custodian discards the packet. When thedestination receives packet, it transmitsacknowledgement to source.LODIS [1] maintains a local database, and updatesit by using the gossiping technique in addition torouting. The database is updated by exchanging theirinformation in database as nodes encounter each other.When the routing protocol requests a location fromLODIS, it provides the location with which the packetsare forwarded to destination and reduces delay. TheLAROD-LODIS maintains a constant overhead for allpossible loads and it has increased delivery ratio [1].Source IP Data IP Destination Power of node Neighbour list TimeoutFigure 6: Data Field of Summary VectorXV. COMPARISION OF ASSORTED PERFORMANCE METRICSTable 1: Comparison of Assorted Performance MetricsS.NOROUTINGTECHNIQUEOVERHEADDELIVERYLATENCYTRANSMISSIONDELIVERYRATE1. Flooding Maximum MaximumMore number oftransmissions-2. Epidemic Maximum Minimum - Maximum
  8. 8. 42 S. Ramesh, P. Ganesh Kumarwww.ijorcs.org3. BLRMinimum (since it’sbased on position)Reduced delayBased on theposition of nodesIndependent ofthe nodemobility4. CAR Minimum Good - Good5. Brownian Gossip MinimumComparativelygoodBased on theprobability PgossipComparativelygood6. SOLAR-HUB Maximum - - Maximum7. D-SOLAR KSP MinimumDecreased athigher userdensityIncreased asnumber of hubsincreases-Increased withhigh userdensityDecreased asnumber of hubsincreases8. S-SOLAR KSP MaximumDecreased athigher userdensityIncreased asnumber of hubsincreases-Increased withhigh userdensityDecreased asnumber of hubsincreases9. DIG Minimum OptimalIncreases as delayincreasesIncreases asqueue sizeincreases.With largequeue size moremessages isbuffered andfewer messagesare dropped.10. Single Copy Minimum GoodBad as copy getslostPoor11. Multiple Copy Maximum OptimalFewertransmissionMaximum12. SPRLess as minimummessage forwarding- - Maximum13.Binary Spray andWait- Good Good -14. Seek and Focus - -Bad as single copygets lost-15. Spray Select Focus Optimal - -Maximum asroute isminimised16. Spray and Hop Highly Scalable Slightly highDecreased numberof transmissionsMaximum17. LAROD-LODISConstant at allnodesOptimalFewertransmissionsMaximum withhigh probability
  9. 9. A Study of Routing Techniques in Intermittently Connected MANETs 43www.ijorcs.orgXVI. CONCLUSION AND FUTURE WORKThis paper depicts the various routing techniquesthat were possibly adapted in ICMANET,Intermittently connected Mobile Ad hoc Network. Italso out frames the performance metrics of variousalgorithms and their comparisons with each other. Thetechniques described in this paper are analyzed in wellversed manner and their comparative evaluation isdepicted in the Table 15.1. These techniques prove thepossibility of routing in the partially connected ad hocnetworks. The study of these routing methodologiesaids in developing new techniques. As a result of thisstudy, it is found that the secure routing in this networkhas not been accomplished. In accordance toenhancement of this study, we would apply securitymeasures to the existing routing protocol and adhere tosecure communication between nodes in the network.XVII. REFERENCES[1] Erik Kuiper and Simin Nadim-Tehrani, “GeographicalRouting with Location Services in IntermittentlyConnected MANETs”, IEEE Trans. Veh. Technol. vol.60, no. 2, pp. 592 – 694, Feb. 2011. doi:10.1109/TVT.2010.2091658[2] T. Spyropoulos, K. Psounis, and C. S. Raghavendra,“Spray and Wait :An Efficient Routing Scheme forIntermittently Connected Mobile Networks” in Proc.ACMSIGCOMM Workshop Delay - TolerantNetworks, 2005, pp. 252-253. doi:10.1145/1080139.1080143[3] W. K. Lai, W. K. Chung, J. B. Tsai, and C. S. Shieh,“Spray and Hop: Efficient Utility–Mobility Routing forIntermittently Connected Mobile Networks” inProceeding of International Conference onCommunication and Networking, Chinacom., 2009.doi: 10.1109/CHINACOM.2009.5339732.[4] E. J. Jebajothi, V. Kavitha, and T. Kavitha, “ContentionBased Routing in Mobile Ad Hoc Networks withMultiple Copies,” in International Journal ofEngineerng. and Technology, vol. 2, 2010, pp. 93-96.[5] K. Shi, “Semi-Probablistic Routing in IntermittentlyConnected Mobile Ad-Hoc Networks” in Journal ofInfo. Science and Engg.,vol. 26, 2010, pp. 1677-1693.[6] T. Spyropoulos, K. Psounis, and C. Ragavendra,“Efficient Routing in Intermittently Connected MobileNetworks: The Multiple-Copy Case,” IEEE/ACMTrans. Netw., vol. 16, no. 1, pp. 77-90, Feb.2008. doi:10.1109/TNET.2007.897964[7] T. Spyropoulos, K. Psounis, and C. S. Ragavendra,“Efficient Routing in Intermittently Connected MobileNetworks: The Single-Copy Case,” IEEE/ACMTransactions on Networking, vol. 16, no. 1, pp. 63–76,Feb. 2008. doi: 10.1109/TNET.20070897962[8] Z. Li, and H. Shen, “A Direction Based GeographicRouting Scheme for Intermittently Connected MobileNetworks,” in IEEE/IFIP International Conference onEmbedded and Ubiquitous Computing, 2008, pp. 359-365. doi: 10.1109/EUC.2008.159,[9] J. Ghosh, H. Q. Ngo, and C. Qiao, “Mobility Profilebased Routing within Intermittently Connected MobileAd Hoc Networks” in Proceedings of ACM WirelessCommunications and Mobile Computing, 2006, pp.551-556. doi: 10.1145/1143549.1143659[10] R. R. Choudhury, “Brownian gossip: Exploiting NodeMobility to Diffuse Information in Ad Hoc Networks,”in Proc. Int. Conf. Collaborative Comput.: Netw., Appl.Worksharing, 2005, pp. 1-5. doi:10.1109/COLCOM.1651262[11] M. Musolesi, S. Hailes, and C. Mascolo, “AdaptiveRouting for Intermittently Connected Mobile Ad HocNetworks” in Proceedings of IEEE 6th InternationalSymposium, WoWMoM, 2005. doi:10.1109/WOWMOM.2005.17[12] M. Heissenbüttel, T. Braun, T. Bernoulli, and M.Wälchi, “BLR: Beaconless Routing Algorithm forMobile Ad Hoc Networks”, ComputerCommunications, vol. 27, no. 11, pp. 1076-1088, July2004. doi: 10.1016/j.comcom.2004.01.012.[13] A. Vahdat and D. Becker, “Epidemic Routing forPartially Connected Ad Hoc Networks,” Duke Univ.,Duhram, NC, TecH.Rep. CS-2000-06, 2000.[14] J. Ghosh, C. Westphal, H. Ngo, and C. Qiao, “BridgingIntermittently Connected Mobile Ad Hoc Networks(ICMAN) with Sociological Orbits”.http://joyghosh.com/docs/INFOCOM_POSTER_SOLAR.pdf[15] D. Cokuslu, K. Erciyes, “A Flooding based RoutingAlgorithm for Mobile Ad Hoc Networks,” in IEEE16th. Int. Conf. SIU 2008, pp. 1-5.http://joyghosh.com/docs/INFOCOM_POSTER_SOLAR.pdf[16] S. Ramesh, P. Ganesh Kumar, “A ComparativeAnalysis of Routing Techniques in IntermittentlyConnected MANETs”, in Asian Journal of ScientificResearch. (In Press).How to citeS. Ramesh, P. Ganesh Kumar, " A Study of Routing Techniques in Intermittently Connected MANETs ".International Journal of Research in Computer Science, 3 (3): pp. 35-43, May 2013. doi: 10.7815/ijorcs.33.2013.066