Dr. P.K.Suri, Professor and Dean Lokesh Pawar, Mteh Student
Department of Computer S. ! "n#ineerin# Department of Computer S. ! "n#ineerin# $.C.T.M, $.C.T.M, Kaithal, $aryana %&ndia' Kaithal, $aryana %&ndia' AbstractDTN (Delay Tolerant Networking) It is a network which is overlaid on the top of regional networks, which also includes the internet. Due to lack of continuous connectivity between the nodes there are several architectures proposed for such an etre!e proposed for such an etre!e and scarce environ!ents for internet. "s DTN approaches towards delay or disruption tolerant network environ!ent which keeps the !essage viable in a scarce networking conditions. This survey eplores various fields of DTN which are being treated so as to !ake DTN viable. This survey tells us about DTN !ultiple fields which are under develop!ent. Keywords ( DTN %Delay Tolerant Networkin#', TCP)&P %Transmission Control Protool ) &nternet Protool', CD %Certifiate *uthority', &D %&dentity', +, %+adio ,re-ueny', *N %*ir.orne Networks' &. &NT+/D0CT&/N Today1s &nternet has .een very suessful at onnetin# ommuniatin# devies round the #lo.e. &t has .een made possi.le .y usin# a set of protools, whih is widely known as TCP)&P protool suite. "very devie on the innumerous su.2networks that omprise the &nternet uses this protool for transferrin# the data from soure to destination with the minimal possi.le delay and hi#h relia.ility. The underlyin# priniple on whih TCP)&P works is .ased on end2to2end data transfer usin# num.er of potentially dissimilar link2layer tehnolo#ies. $owever, there are many re#ions where the assumptions of the internet annot .e upheld. &f at any instant there is no path .etween the soures to destination, then TCP)&P fails to work properly or mi#ht even stop workin# ompletely. 3eause of suh irumstanes, a newer network has evolved whih is independent end to end onnetivity .etween nodes. This network is alled as DTN. Delay Tolerant Networkin# is an approah to omputer network arhiteture that aims to address the tehnial issues in hetero#eneous networks that e4periene lak of ontinuous network onnetivity. DTNs ena.le data transfer when mo.ile nodes are only intermittently onneted. Due to lak of onsistent onnetivity, DTN routin# usually follows store2arry2and2forward5 i.e., after reeivin# some pakets, a node arries them around until it ontats another node and then forwards the pakets. Sine DTN routin# relies on mo.ile nodes to forward pakets for eah other, the routin# performane %e.#., the num.er of pakets delivered to their destinations' depends on whether the nodes ome in ontat with eah other or not. A. CONCEPT of DTN * Delay Tolerant Network an .e onsidered as an overlay on the e4istin# re#ional networks. This overlay is alled as the .undle layer. This layer is intended to funtion a.ove the e4istin# protool layers and provide the funtion of a #ateway when two nodes ome in ontat with eah other. The main advanta#e of this kind of protool is fle4i.ility. &t an .e easily linked with the already e4istin# TCP)&P protool networks or an .e used to link two or more networks to#ether. The position of the .undle layer an .e seen in the followin# fi#. 6.
,i# 67 The position of the .undle layer. 3undles are also alled as messa#es. The transfer of data from one node to another an .e made relia.le .y storin# and forwardin# entire .undles .etween nodes. The .undles omprise of three thin#s, "pplication #ayer "pplication #ayer "pplication #ayer $%ND#& #"'&( (egion )pecific #ayers (egion )pecific #ayers (egion )pecific #ayers soure node1s user2data, ontrol information %e.#., soure node &D, destination node &D, TTL et.', a .undle header. 3esides 3undle transfer, ustody transfer is also done. The ustodian node for a .undle keeps the messa#e until it is suessfully transferred to the ne4t node and it takes the ustody for that messa#e or until the TTL of the messa#e e4pires. $. $%ND#& #"'&( $undle *reation at )ource The soure appliation invokes its .undle layer, re-uestin# transfer of a .undle with a header as shown in the ta.le .elow. The soure1s user data inludes instrutions to the destination appliation for proessin#, stora#e, disposal, and error2handlin# of the data. This user data is not visi.le to the .undle layers handlin# the transfer. The soure .undle layer verifies the soure1s si#nature, reates a .undle, appends its own si#nature after the .undle header, and stores the result in persistent stora#e. The stora#e is re-uired, even if an immediate forwardin# opportunity e4ists, .eause the .undle layer has aepted a ustody transfer and must therefore .e prepared to retransmit the .undle if it does not reeive aknowled#ement, within the .undle1s time2 to2aknowled#e, that the su.se-uent ustodian has reeived and aepted the .undle. Trans!ission by )ource The soure .undle layer onsults its routin# ta.le and finds that the ne4t hop apa.le of aeptin# ustody transfers on a path toward the destination, and that TCP is the proper transport protool. The soure .undle layer also determines that it has a ontinuous onnetion to the "arth #ateway. The .undle layer transmits a opy of the .undle to the ne4t #ateway via TCP, starts a time2to2aknowled#e retransmission timer, and awaits a ustody2transfer aknowled#ment from the #ateway. $undle (eception by Destination 8hen the destination .undle layer reeives the .undle via TCP, it terminates the TCP session and heks the si#nature of the Mars2#ateway .undle layer, usin# its stored opies of ad9aent2router ertifiates and ertifiate2authority %C*' pu.li keys. &t determines that the .undle has .een forwarded .y a le#itimate soure. Then it stores the reeived .undle in persistent stora#e, aepts ustody of the .undle, and onfirms this .y aknowled#ement to the senders .undle layer, whih deletes its ustodial opy of the .undle. The destination .undle layer awakens the destination appliation identified .y the entity &D. Dependin# on the ontrol part of the user data sent .y the soure, the destination appliation may #enerate an appliation2layer aknowled#ment in a new .undle and send it to the soure. *. )T+(&,*"((',"ND,-+(."(D Delay Tolerant Networks have overome the pro.lems assoiated with the onventional protools in terms of lak of onnetivity, irre#ular delays, asymmetri .idiretional data rates et. usin# the onept of store arry and forward. The method of store and forward is very analo#ous to the real life postal servie. "ver y letter has to pass throu#h a set of post offies, where it is proessed and forwarded, .efore reahin# the destination. $ere the omplete messa#e or a hunk of it is transferred and stored in nodes suessively until reahes the destination. The followin# fi#ure %fi#. :', #ives a rou#h #raphial representation of how a messa#e is propa#ated throu#h a network. ,i# : 7 Store and forward approah in DTN layers "ah node is assoiated with a persistent stora#e devie %like hard disk', where it an store the messa#es. &t is alled as persistent stora#e as it an store the messa#e for indefinite amount of time unlike short2term memory devies. The persistent stora#e an .e useful in situations when the ne4t node is not availa.le for a very lon# time, or when the rate of inomin# messa#es is far hi#her than the rate of out#oin# messa#es. D. DTN "//#I*"TI+N) 0. &!ergency )ervices Searh and resue operations Disaster reovery +eplaement of fi4ed infrastruture in ase of environmental disasters Poliin# and fire fi#htin# Supportin# dotors and nurses in hospitals Node $ Node * Node " 1. )ensor Networks $ome appliations7 smart sensors and atuators em.edded in onsumer eletronis 3ody area networks %3*N' Data trakin# of environmental onditions, animal movements, hemial).iolo#ial detetion 2. Terrestrial 3obile Networks 4. &otic 3edia Networks Near "arth Satellite ;ery Lon# Distane +adio Links %Deep Spae +,' 5. 3ilitary "d,6oc Network &. N&&D -+( DTN These networks are harateri<ed .y the followin#. &t is .eause of these harateristis that &nternet Protools fail or is rendered useless. 6'Lak of Connetivity7 &f at any moment, there is no end2to2end path .etween soure and destination %widely alled network partitionin#', then end2to2end ommuniation annot take plae usin# the TCP)&P protools suite. $ere DTN omes very useful. :' &rre#ular Delays7 Lon# delays an ause the TCP)&P protool suite to funtion improperly. Propa#ation delays .etween transmittin# nodes ompounded with -ueuin# delay at eah node an topple the protools whih rely lar#ely on -uik return of aknowled#ement of a sent data. This an .e overome usin# DTNs. =6> ?' *symmetri 3idiretional Data +ates7 Moderate asymmetries of .idiretional data rate an .e tolerated to an e4tent in onventional protools. 3ut if asymmetries are lar#e, they an .e defeated easily. Networks alled as hallen#ed networks violate the assumptions of the onventional &nternet and hene TCP)&P protools an1t .e used here. *s desri.ed in =6> the e4amples of hallen#ed networks an .e "4oti Media Networks, Terrestrial Mo.ile Networks, Sensor2.ased Networks et. && S0+;"@ /N DTN (eal .orld Deploy!ent Delay2 and Disruption2 Tolerant Networkin# %DTN' has .een a researh topi for more than 6A years sine ;inton Cerf and N*S* started workin# on means to e4pand &nternet2 like onnetivity to situations suh as ommuniations with spaeraft on deep spae missions. The lar#e round trip delays make onventional &nternet protools suh as TCP unusa.le. The DTN arhiteture and the 3undle Protool suite that have .een developed to ena.le ommuniation in hallen#ed environments, .oth in spae and in terrestrial situations, have now reahed a level of development and testin# where they are a.le to .e used to support users in real2world situations. "lwyn Davies desri.es some of the work that has .een done to demonstrate real2world appliations of DTN, and how DTN an .e used to support a future &nternet where information items .eome Bfirst lass iti<ensB, ena.lin# &nformation Centri Networkin# %&CN'. =C> (outing in socially selfish DTN tells us a.out the different sheme for routin# in DTN from e4istin# inentive .ased shemes. These new shemes stimulate individually selfish nodes to forward for all other nodes they followed a new philosophy of Ddesi#n for userE=?>. They took soial selfishness as a user demand and allow soially selfish nodes to .ehave in the aforementioned ways to satisfy suh demand. There are selfish nodes whih are usually willin# to forward pakets to other nodes with whom they have soial ties .ut not to others.They proposed a Soial Selfishness *ware +outin# %SS*+' al#orithm to maintain the routin# performane and to maintain the soial selfishness. SS*+ al#orithm onsiders two points one is the willin#ness to forward and other is their ontat opportunity whih results in a .etter forwardin# strate#y than purely ontat .ased approah. SS*+ allows users to maintain selfishness and ahieves .etter routin# performane with low transmission ost =?>. )ervice Discovery * novel servie disovery protool for delay tolerant networks in whih all node onnetions are intermittent. Most of the previously proposed protools are .ased on the assumption that there e4ists an end2to2end onnetion from the -uery soure node to the destination node, the assumption that rarely holds for mo.ile wireless networks. The desi#n of servie disovery protools is partiularly hallen#in# for mo.ile wireless networks .eause of their dynami and unstrutured nature. To desri.e servies throu#h a fi4ed si<e strin#, they used a 3loom filter that leads to effiient servie announement and searh. There servie disovery protool is a diretory2less method, whih uses periodial .roadast to announe servie advertisements =F>. To redue servie advertisement paket si<e and memory usa#e, 3loom filter is used to represent servies with a fi4ed si<e strin#. The servie -uery pakets are sprayed into the network and relayed to nei#h.or nodes that meet the destination node providin# the desired servie reently. They used a tehni-ue .ased on 3loom filters =G> =6A> to desri.e servies. The spray and wait routin# al#orithm =H> and random walk .ased searh method =66> are om.ined to forward servie -uery and servie data pakets in =I> =J>, history of last meetin#s .etween nodes is used as hints to aid the paket forwardin#, .ut we used this information in a different way. 3ulticasting in DTN Multiast in DTNs with sin#le and multiple data items, investi#ated the essential differene .etween multiast and uniast in DTNs, and formulated relay seletions for multiast as a unified knapsak pro.lem .y e4ploitin# node entrality and soial ommunity strutures. "4tensive trae2driven simulations show that our approah has similar delivery ratio and delay to the "pidemi routin#, .ut an si#nifiantly redue the data forwardin# ost measured .y the num.er of relays used =6:>. They foused on improvin# the ost2 effetiveness of multiast in DTNs .y e4ploitin# the two key onepts in Soial Network *nalysis, i.e., entrality and ommunities. The .asi idea of their approah is to develop soial2.ased metris .ased on the pro.a.ilities of the nodes forwardin# data to their destinations. The metris in =6:> were developed .ased on soial network onepts inludin# entrality and soial ommunities. They investi#ated the essential differene .etween multiast and uniast in DTNs, and developed relay seletion shemes onsiderin# the forwardin# pro.a.ilities to multiple destinations simultaneously. 3ultiregion DTN ,or multire#ion delay tolerant networked systems many appliation senarios an .e envisioned. ,or e4ample, let us onsider the lassi e4ample of users wishin# to send email from a mo.ile devie or a laptop %not onneted to any &nternet aess point' to a user in the &nternet =6?>. Messa#es should .e routed firstly to &nternet #ateways and then forwarded in the usual way throu#h the &nternet. &n other words, the routin# proess onsists in findin# an inter2re#ion #ateway that onnets the two re#ions. *lmost all the e4istin# work on routin# in delay tolerant networks has foused on the pro.lem of delivery of messa#es inside a sin#le re#ion, harateri<ed .y the same network infrastruture and namespae. They introdued a proposal for inter2re#ion routin# .ased on .oth pro.a.ilisti and deterministi forwardin# mehanisms, em.edded in an arhitetural framework a.le to support it. Multi2re#ion routin# shemes an also .e onsidered of key importane for desi#nin# systems to support healthare initiatives. Many systems .ased on mo.ile phones for remote healthare interventions have .een presented in the reent years =6?>. They also introdued a model of the pro.lem and deompose it into intra2 re#ion routin# %i.e., routin# inside a re#ion' and inter2 re#ion routin# %i.e., routin# aross re#ions'. Then they desri.ed how to use delivery pro.a.ilities to implement the forwardin# proess inside a re#ion and also disussed the namin# system in N/M*D =6?>. /ro!oting Tolerance for DTN to address some of the omplaints a.out DTN. The authors present su##estions of e4petations for appliations, and metris for performane, whih su##est a more tolerant view of researh in the area of Disruption Tolerant Networkin#. They have already pointed out, there was a lar#e #ap .etween the development of the &nternet arhiteture and the first ompellin# appliations. This does not mean, however, that we should, or even an, i#nore the appliations in =:J> that mi#ht .enefit from a DTN arhiteture. They were not a.le to laim That they have predited the killer appliations for DTNs, instead, They desri.ed some of the aspets of DTNs that an ena.le novel appliations. The apaity of DTN will always .e muh hi#her than in an infrastruture network =6C>.This artile has attempted to demonstrate that DTNs have the potential to ena.le e4itin# new appliations that annot .e served .y the urrent &nternet. "pplications of DTN in "N the urrent air.orne network is omposed of multiple platforms whih ommuniate aross numerous data links and protools. &n the interim timeframe, /.9etive Kateway %/K' platforms suh as the 3attlefield *ir.orne Communiations Node %3*CN' and +apid *ttak &nformation Dissemination "4eution +elay %+*&D"+' may provide interim interopera.ility solutions in the *N environment throu#h waveform and media translation servies, however widespread deployment has yet to .e determined =6H>. *s *ir.orne Networks %*Ns' evolve and .eome inreasin#ly net2entri, it is ritial that apt onsideration is #iven to properly e4tend &nternet Protool %&P'2.ased networkin# servies to the tatial assets. *uthors said that this is an effort to resolve urrent interopera.ility issues throu#h the use of a standard protool and provide &P onnetivity .etween all *ir.orne Network platforms. =6H> addresses the speifi hallen#es of esta.lishin# and maintainin# relia.le ommuniations within an &P2 .ased *N environment. &n addition, =6H> provides an overview of DTN, and details how this tehnolo#y may .e inte#rated into to the *N infrastruture in order to address the urrent issues affetin# *N performane. Distributed *o!!unity Detection this is a reasona.le intuition, sine people in the same ommunity are likely to meet re#ularly and hene .e appropriate forwarders for messa#es destined for other mem.ers of their ommunity. Similarly, in the area of soiolo#y, the idea of Dorrelated interationE is that an or#anism of a #iven type is more likely to interat with another or#anism of the same type than with a randomly hosen mem.er of the population. Community is an important attri.ute of Poket Swithed Networks %PSN', .eause mo.ile devies are arried .y people who tend to .elon# to ommunities. They proposed and evaluated three novel distri.uted ommunity detetion approahes with #reat potential to detet .oth stati and temporal ommunities. They found that with suita.le onfi#uration of the threshold values, the distri.uted ommunity detetion an appro4imate their orrespondin# entrali<ed methods up to GAL auray.=6F> (outing for Developing regions The pro.lem of routin# in delay tolerant networks deployed in developin# re#ions. *lthou#h these environments e4periene intermittent onnetivity %hene the desire to use DTN', in many ases the topolo#y has an underlyin# sta.ility that we an e4ploit when desi#nin# routin# protools. 3y makin# small, yet ritial, modifiations to lassial link state routin#, we derive a more effetive al#orithm apa.le of levera#in# preditions of future link uptimes =6I>. They desri.ed a omplete and fully2implemented protool, apa.le of .ein# deployed in the DTN referene implementation without modifiation. 3y usin# a simulation in =6I> inorporatin# real2world network harateristis, authors demonstrated that their system operates effetively when onventional routin# and forwardin# may fail. 3odelling (edundancy $ased (outing +outin# in delay tolerant networks is hallen#in# .eause many well2known assumptions of traditional networks do not hold in DTNs. ;arious routin# shemes have .een proposed in literature whih e4ploit redundany to improve the messa#e delivery. $owever, there laks a systematial analyti work on modelin# the redundany .ased routin# shemes for delay tolerant networks. They model the messa#e deliverin# proess in DTNs as a ontinuous time Markov random proess with a.sor.in# state. 8ith this model, they have derived the pro.a.ility distri.ution funtions of the messa#e delivery delay for different routin# shemes, inludin# diret transmission routin#, soure forward, and .inary forward shemes. 8e ondut simulations to validate our model and our analyti results. &n order to ontrol the overhead of floodin#2.ased routin# shemes, the authors of =6J> proposed to dispath a ertain num.er of idential messa#e opies to a fi4ed num.er of relay nodes, instead of sendin# to every node in the network. 8hen a messa#e is #enerated .y a soure node, a D-uotaE is attahed to that messa#e, whih represents how many idential opies of the messa#e an .e inserted in the network. "nycast (outing 3ased on a new DTN model, they first analy<ed the anyast semantis for DTNs. * novel metri named "MDD* was presented in =6G>. 8ith orrespondin# routin# al#orithm for anyast routin# in DTN. "MDD* %"4peted Multi2 Destination Delay for *nyast' routin# sheme effetively improved the effiieny of anyast routin# in DTNs . *nyast is a servie that allows a node to send a messa#e to at least one, and prefera.ly only one, of the mem.ers in a #roup. The idea .ehind anyast is that a lient wants to send pakets to any one of several possi.le servers offerin# a partiular servie or appliation .ut does not really are any speifi one. *nyast an .e used to implement resoure disovery mehanisms whih are powerful .uildin# .loks for many distri.uted systems, inludin# file sharin# et. )ecurity *onsiderations &nternet2inspired seurity work on delay tolerant networkin#, in partiular, as it mi#ht apply to spae missions, and identifies some hallen#es arisin#, for .oth the &nternet seurity ommunity and for spae missions. These hallen#es inlude the development of key mana#ement shemes suited for spae missions as well as a harateri<ation of the atual seurity re-uirements applyin#. * speifi #oal of this paper is therefore to eliit feed.ak from spae mission &T speialists in order to #uide the development of seurity mehanisms for delay tolerant networkin#.=:A> /roble! of 3ulticasting The pro.lem of multiastin# was first studied in =:6> to the .est of our knowled#e. &t supports the distri.ution of data to a #roup of users a servie needed for many DTN appliations =:6>. Multiastin# in DTN is a onsidera.ly different and hallen#in# pro.lem =:6>. &n =:6> there are several multiast routin# al#orithms developed with different routin# strate#ies. Multiastin# is a hallen#in# task in DTN .eause there is no end2to2end path .etween some or all nodes.The DTN network have a variety of appliations in situations that inlude i'.risis environments like emer#eny response and military .attlefields, deep2spae ommuniation, vehiular ommuniation and non2interative &nternet aess in rural areas =:6>. Multiastin# in the &nternet and Manet has .een studied e4tensively in the past =:6>. Three multiast semanti models were developed whih allows users to impliitly speify the temporal onstraints on #roup mem.ership and messa#e delivery .,our lasses of routin# al#orithms for DTNs with different routin# strate#ies has .een developed in =?F> and how routin# performanes in effeted .y the availa.ility of knowled#e. /ractical (outing DTN have the potential to onnet devies in hallen#ed environments .y the urrent networks. There are several hallen#es for DTN suh as determinin# routes throu#h the network ever havin# an end2to2end ommuniation =::>. The approahes prior to this have .een foused on epidemi messa#e repliation or on the knowled#e of the onnetivity shedule =::>. "pidemi routin# approah of repliation of messa#es is e4pensive and it doesn1t appear to sale well with inreasin# load. * metri has .een desi#ned that estimates how lon# a messa#e will have to wait .efore it an .e transferred to ne4t hop =::>. &n =::> the topolo#y is distri.uted .y usin# link2state routin# protool and the link2state pakets are DfloodedE usin# epidemi routin#. The link2state routin# has .een implemented here .eause it provides the omplete topolo#y at eah node. * framework for evaluatin# DTN routin# al#orithms has .een provided in =:?>.The al#orithms whih do well in these environments are mentioned in =:?>. The per2ontat term learly shows a win situation for DTN whereas soure and per2hop routin# drop the messa#e. The routin# ta.les are reomputed every time a onnetion is made =:?>. (outing in DTN in a DTN, an end2to2end path may .e unavaila.le at all times routin# is performed over time to ahieve eventual delivery .y employin# lon#2 term stora#e at the intermediate nodes =:?>.The formulation and motivation of DTN routin# pro.lem when the onnetivity patterns are known then provide a framework for evaluatin# various routin# al#orithm =:?>. DTN routin# appears to .e a rih and hallen#in# pro.lem. &t re-uires tehni-ues to selet paths shedule transmissions, estimate delivery performane and mana#e .uffers. &n those situations where resoures are limited %ontat opportunities, .andwidth or stora#e' smarter al#orithms may provide a si#nifiant .enefit. DTN "rchitecture There were several pro.lems assoiated with M*N"T. To resolve these pro.lems an arhiteture in =:C> has .een developed to ahieve more interopera.ility .etween e4treme environments lakin# ontinuous onnetivity. Many of suh networks have their own speiali<ed protools and do not utili<e &P. The DTN arhiteture aims to provide interopera.le ommuniations .etween a wide ran#e of networks whih may have e4eptionally poor and disparate performane harateristis. &&&. Conlusion 8hen there is a lak of onnetivity or intermittent onnetivity DTN supports in these environments. DTN supports lon# and varia.le delays. &t uses a store2arry and forward approah whih onsists of arryin# a messa#e on the node until the nodes #ets onneted to the other node desired. 8e have disussed papers related to arhiteture, Multiastin#, *nyastin#, Seurity, +outin# in DTN. +eferenes 6. ,.8arthman. Delay Tolerant Networks ( * Tutorial. :AA?. =:> ;. Cerf, S. 3urlei#h, *. $ooke, L. Tor#erson, +. Durst, K. Sott, K. ,all, $. 8eiss,Delay2Tolerant Network *rhiteture, DTN +esearh Kroup &nt ernet Draft, :AA?. :. ;. Cerf, S. 3urlei#h, *. $ooke, L. Tor#erson, +. Durst, K. Sott, K.,all, $. 8eiss, Delay Tolerant Network *rhiteture, DTN +esearh Kroup &nternet Draft, Draft :, Marh :AA?. ?. > M. Li, S. Nhu, K. Cao. +outin# in Soially Selfish Delay Tolerant Networks. &n proeedin#s of &""" &nfoom, :A6A. C. "lwyn Davies. Delay2Tolerant Networkin#7 Movin# Towards +eal28orld Deployment. Presented at &nformation Soiety :A66,L9u.l9ana. H. Thrasyvoulos Spyropoulos, Konstant inos Psounis, and Cauli#i S. +a#havendra. Spray and wait7 *n effiient routin# sheme for intermittently onneted mo.ile net works. &n 8DTN :AAH7 Proeedin# of the :AAH *CM S&KC/MM workshop on Delay2tolerant networkin#, :AAH. F. Ni9ian 8an#, "yuphan 3ulut, and 3oleslaw K. S<ymanski. DServie Disovery for Delay Tolerant Networks,E *ppears in KL/3"C/M 8orkshops %KC 8kshps', :A6A &""". I. *. Lind#ren, *. Doria, and /. Shelen, Pro.a.i listi routin# in intermittently onneted networks, Mo.i$o1 A?,Oune 62 ?,:AA?,*nnapolis, Maryland,0S*. J. 3. Chen, C. Choon, Mo.iCent7 a redit2.ased inentive system for disruption tolerant network, &n Pro. &""" &N,/C/M, :A6A. S. Nhon#, O. Chen, @.+. @an#. Sprite7 a simple, heat2proof, redit2.ased system f or mo.ile ad2ho networks, &n Pro. &""" &N,/C/M, :AA?. G. &. Sheriff, P. *. K. *harya, *. Sampath, 3.@. Nhao and ". M.3eldin#, D&nte#rated data loation in multihop wireless networks,E Proeedin#s of the seond &nternational Conferene on Communiation Systems Software and Middleware, pp. 626A, :AAI 6A. L. ,an, P. Cao, O. *lmeida, and *. 3roder, DSummary ahe7 * sala.le wide2area 8e. ahe sharin# protool,E in Pro. of *CM S&KC/MM Conf., Sept. 6GGJ, pp. :HC(:FH. 66. *. N. Mian, +. 3eraldi and +. 3aldoni, D&dentifyin# open pro.lems in random walk .ased servie disovery in mo.ile ad ho networks,E Proeedin#s of the Fth &nternational 8orkshop on &nnovative &nternet Community Systems, :AAF. 6:. 8ei Kao, Min#hua Li, 3o Nhao and Kuohon# Cao. DMultiastin# in Delay Tolerant Networks7 * Soial Network Perspetive,E Mo.i$o1AG,May 6J2:6,:AAG, New /rleans, 0S*. 6?. Miro Musolesi ,Ceilia Masolo D* ,ramework for Multire#ion Delay Tolerant Networkin#,E 8iNS2D+1AJ,Septem.er 6G,:AAJ,San ,raniso,California,0S*. 6C. Kross#lauser, M., and Tse, D. N. C. Mo.ility inreases the apaity of ad ho wireless networks. &""")*CM Transations on Networkin# 6A, C %*u#. :AA:', CII(CJF. 6H. Thomas Oonson, Oonah Pe<eshki, ;itor Chao, Kristofer Smith, Oames ,a<io 3oo< *llen $amilton $erndon, ;*. *ppliation of Delay Tolerant Networkin# %dtn' in *ir.orne Networks Military Communiations Conferene, :AAJ. M&LC/M :AAJ. &""". 6F. Pan $ui, "iko @oneki, Shu2@an Chan, Oon Crowroft. Distri.uted Community Detetion in Delay Tolerant Networks. Mo.i*rh1AI, *u#ust :I(?6, :AAI, Kyoto, Oapan. 6I. Mihael Demmer, Kevin ,all. DTLS+7 Delay Tolerant +outin# for Developin# +e#ions. NSD+1AI, *u#ust :I, :AAI, Kyoto, Oapan. 6J. @on# Liao, Kun Tan , Nhenshen# Nhan#, Li4in Kao. Modelin# +edundany2.ased +outin# in Delay Tolerant Networks. Consumer Communiations and Networkin# Conferene, :AAI. CCNC :AAI. Cth &""". 6G. @ili Kon#P, @on#-ian# Qion#, Mian Nhan#, Nhenshen# Nhan#, 8en9ie 8an#, Nhiwei Qu. *nyast +outin# in Delay Tolerant Networks. Mirosoft +esearh Teh+eport Num.er7MS+2 T+2:AAF2AC. :A. Stephen ,arrell, ;inny Cahill. Seurity Considerations in Spae and Delay Tolerant Networks. :nd &""" &nternational Conferene on Spae Mission Challen#es for &nformation Tehnolo#y %SMC2&TBAF'. :6. 8enrui Nhao, Mostafa *mmar and "llen Ne#ura. Multiastin# in Delay Tolerant Networks7 Semanti Models and +outin# *l#orithms. S&KC/MMBAH 8orkshops, *u#ust ::(:F, :AAH, Philadelphia, P*, 0S*. ::. "van P. C. Oones Lily Li and Paul *. S. 8ard. Pratial +outin# in Delay2Tolerant Networks. S&KC/MM1AH 8orkshops, *u#ust ::(:F, :AAH, Philadelphia, P*, 0S*. :?. Sushant Oain Kevin ,all and +a.in Patra. +outin# in a Delay Tolerant Network. S&KC/MM1AC, *u#. ?A(Sept. ?, :AAC, Portland, /re#on, 0S*. :C. Kevin ,all. * Delay2Tolerant Network *rhiteture for Challen#ed &nternets. S&KC/MM1A?, *u#ust :H2:G, :AA?, Karlsruhe, Kermany.