ALM-FastReplica: Optimizing the Reliable Distribution of Large Files within CDNs

Size: px
Start display at page:

Download "ALM-FastReplica: Optimizing the Reliable Distribution of Large Files within CDNs"

Transcription

1 ALM-FastReplca: Optzng the Relable Dstrbuton of Large Fles wthn CDNs Ludla Cherkasova Internet Systes and Storage Laboratory HP Laboratores Palo Alto HPL Aprl 4, 005* E-al: CDNs, edge servers, orgn server, relablty, large fle replcaton, content delvery, perforance analyss, scalablty Content Delvery Networks (CDNs) provde an effcent support for servng http and streang eda content whle nzng the network pact of content delvery as well as overcong the server overload proble. For servng the large docuents, software download packages and eda fles, there s an addtonal proble of the orgnal content dstrbuton across the edge servers n CDN nfrastructure. We propose an algorth, called ALM-FastReplca, for optzng replcaton of large fles across the edge servers n CDNs. The orgnal fle s parttoned nto k subfles, and each sub fle s replcated va a correspondngly constructed ultcast tree. Nodes fro the dfferent ultcast trees use addtonal cross nodes connectons to exchange ther correspondng sub fles such that each node eventually receves an entre fle. Ths new replcaton ethod sgnfcantly reduces fle replcaton te, up to 5-5 tes copared to the tradtonal uncast (or pont-to-pont) schea. Snce a sngle node falure n the ultcast tree durng the fle dstrbuton ay pact the fle delvery to a sgnfcant nuber of nodes, t s portant to desgn an algorth whch s able to deal wth node falures. We augent ALM-FastReplca wth an effcent relablty echans, that can deal wth node falures by akng local repar decsons wthn a partcular replcaton group of nodes. Under the proposed algorth, the load of the faled node s shared aong the nodes of the correspondng replcaton group, akng the perforance degradaton gradual. * Internal Accesson Date Only A slghtly shorter paper verson wll be publshed n Proceedngs of the 0th IEEE Syposu on Coputers and Councatons (ISCC'05), 7-30 June 005, La Manga del Mar Menor, Cartagena, Span Approved for External Publcaton Copyrght 005 Hewlett-Packard Developent Copany, L.P.

2 ALM-FastReplca: Optzng the Relable Dstrbuton of Large Fles wthn CDNs Ludla Cherkasova Hewlett-Packard Laboratores 50 Page Mll Road, Palo Alto, CA Abstract Content Delvery Networks (CDNs) provde an effcent support for servng http and streang eda content whle nzng the network pact of content delvery as well as overcong the server overload proble. For servng the large docuents, software download packages and eda fles, there s an addtonal proble of the orgnal content dstrbuton across the edge servers n CDN nfrastructure. We propose an algorth, called ALM-FastReplca, for optzng replcaton of large fles across the edge servers n CDNs. The orgnal fle s parttoned nto k subfles, and each subfle s replcated va a correspondngly constructed ultcast tree. Nodes fro the dfferent ultcast trees use addtonal crossnodes connectons to exchange ther correspondng subfles such that each node eventually receves an entre fle. Ths new replcaton ethod sgnfcantly reduces fle replcaton te, up to 5-5 tes copared to the tradtonal uncast (or pont-to-pont) schea. Snce a sngle node falure n the ultcast tree durng the fle dstrbuton ay pact the fle delvery to a sgnfcant nuber of nodes, t s portant to desgn an algorth whch s able to deal wth node falures. We augent ALM-FastReplca wth an effcent relablty echans, that can deal wth node falures by akng local repar decsons wthn a partcular replcaton group of nodes. Under the proposed algorth, the load of the faled node s shared aong the nodes of the correspondng replcaton group, akng the perforance degradaton gradual. Introducton The aount of streang eda content onlne ncrease everyday. Maor news copanes are aong the largest producers of streang eda content. Sports and usc vdeo stes add ther lon share to popular eda content. Vdeo clps of portant events (SuperBowl, Olypcs, Electon 000, Electon 004) drew unprecedented aounts of traffc. Streang eda can easly becoe a bandwdth hog that overwhels networks and pacts ssoncrtcal applcatons unless the specal content dstrbuton technques are used. Fast and relable dstrbuton of data fro a sngle source to a large nuber of recevers located across the Internet has got a lot of attenton durng last decade. Content Delvery Networks (CDNs) (e.g. Akaa), eploy a dedcated set of achnes (edge servers) to dstrbute content to clents on behalf of the orgn server. In ths soluton, content s replcated at the edge servers (typcally through cachng) closer to the end user, thereby provng content delvery latency whle reducng WAN traffc. For eda fles, t s desrable to replcate these fles at A slghtly shorter paper verson wll be publshed n Proceedng of the 0th IEEE Syposu on Coputers and Councatons (ISCC 005). edge servers n advance, usng the so-called push odel. For large eda fles t s a challengng, resource-ntensve proble, e.g. eda fles can requre sgnfcant bandwdth and download te due to ther large szes: 0 n eda fle encoded at Mbt/s results n a fle of 50 MBytes. Whle transferrng a large fle wth ndvdual pont-to-pont connectons fro an orgnal server can be a vable soluton n the case of lted nuber of rror servers (tenths of servers), ths ethod does not scale when the content needs to be replcated across a CDN wth thousands of geographcally dstrbuted achnes. In our recent work [6], we proposed a new algorth FastReplca for replcatng large fles n the Internet envronent. There are a few basc deas exploted n FastReplca. In order to replcate a large fle aong k nodes (k s n the range of 0-30 nodes), the orgnal fle s parttoned nto k subfles of equal sze and each subfle s transferred to a dfferent node n the group. After that, each node sends ts subfle to the reanng nodes n the group, and the replcaton process s teratvely repeated by takng the nodes wth already replcated fle as the orgn nodes. A fle transfer n FastReplca follows a store-and-forward ode,.e. only after a subfle s receved by a node, ths node starts sendng the correspondng subfle to the subsequent nodes. In ths store-and-forward odel, when the node falure occurs, t ay only pact the content delvery for nodes n the local replcaton group. The recovery algorth s soewhat natural and straghtforward: t only nvolves the nodes wthn the correspondng replcaton group. Snce the natve IP ultcast has not receved wdespread deployent, any ndustral and research efforts shfted to nvestgatng and deployng the applcaton level ultcast, where nodes across the Internet act as nteredate routers to effcently dstrbute content along a predefned esh or tree. A growng nuber of researchers [4, 5, 8,, 4, 9] have advocated ths alternatve approach, where all ultcast related functonalty, ncludng group anageent and packet replcaton, s pleented at end systes. In ths work, we propose to cobne FastReplca deas wth applcaton level ultcast (ALM) for optzng the replcaton of the large fles wthn CDN. We call ths

3 ethod ALM-FastReplca. The orgnal fle s parttoned nto k subfles of equal sze, and each subfle s replcated va a correspondngly constructed ultcast tree usng a fastforward ode,.e. each node starts a fle transfer to the next nodes edately after recevng the frst packet of the correspondng subfle. Nodes fro the dfferent ultcast trees use addtonal cross-nodes connectons to exchange ther correspondng subfles such that each node eventually receves an entre fle. Ths new replcaton ethod sgnfcantly reduces fle replcaton te, up to 5-5 tes copared to the tradtonal ultple uncast ethod. In ultple uncast ethod, the whole fle s transferred n a pont-topont fashon to the frst group of nodes, and after that each node fro the frst group ntates a fle replcaton process to the next group, etc. ALM-FastReplca explots dverse set of Internet paths wthn the replcaton groups to exchange the copleentary content between these nodes. Multple Descrpton Codng (MDC) codes a eda strea nto ultple copleentary descrptons [8, 0, 5]. These descrptons have the property that f ether descrpton s receved t can be used to decode the baselne qualty vdeo, and ultple descrptons can be used to decode proved qualty vdeo. We show how to apply the ALM-FastReplca algorth for an effcent replcaton of ulteda fles encoded wth Multple Descrpton Codng (MDC). Ths new replcaton ethod sgnfcantly reduces the dstrbuton te of MDC-encoded fles to the edge servers n CDN. Under the store-and-forward odel, the node falure ay only pact the content delvery n the local replcaton group. However, under the fast-forward odel, a sngle node falure n a ultcast tree durng the fle dstrbuton ay pact the fle delvery to a sgnfcant nuber of nodes. Thus, t s portant to desgn an algorth whch s able to deal wth node falures. It s a non-trval task because each node n the ultcast tree has a wde fan-out (0 or ore connectons out). A soluton that s based on a straghtforward approach, where the tree below the faled node N faled s reconnected to ts predecessor-node N prev, wll suffer fro poor perforance. Under such a soluton, node N prev becoes a severe bottleneck n the whole syste snce t needs to handle the doubled nuber of connectons. We augent ALM-FastReplca wth an effcent relablty echans, that can deal wth node falures by akng local repar decsons wthn a partcular replcaton group of nodes. Under the proposed algorth, the load of the faled node s shared aong the nodes of the correspondng replcaton group, akng the perforance degradaton practcally unnotceable. The rest of the paper explans these deas n ore detal. Related Work Durng few last years, fast and relable dstrbuton of data fro a sngle source to a large nuber of recevers located across the Internet has been a topc of any research studes. One ethod conssts of accessng ultple servers n parallel to reduce downloadng te or to acheve fault tolerance. The authors n [6], deonstrate proved response te observed by the clent for a large fle download through the dynac parallel access schea to replcated content at rror servers. Dgtal Fountan [] apples Tornado codes to acheve a relable data download. In ther subsequent work [3], the download tes are reduced by havng clent receve a Tornado encoded fle fro ultple rror servers. The target applcaton under ther approach s bulk data transfer. In recent years, overlay networks have becoe an effectve alternatve to IP ultcast for effcent pont to ultpont councaton across the Internet. An nterestng extenson for the end-syste ultcast s ntroduced n [4], where authors, nstead of usng the end systes as routers forwardng the packets, propose that the end-systes do actvely collaborate n nfored anner to prove the perforance of large fle dstrbuton. The an dea s to overcoe the ltaton of the tradtonal servce odels based on tree topologes where the transfer rate to the clent s defned by the bandwdth of the bottleneck lnk of the path fro the server. The authors propose to use addtonal cross-connectons between the end-systes to exchange the copleentary content these nodes have already receved. However, the authors do not explctly partton the orgnal docuent nto a set of copleentary subfles for speedng up the fle delvery. Ther ethod reles on the assupton that any gven par of end-systes has not receved exactly the sae content, and these cross-connectons between the end-systes can be used to reconcle the dfferences n receved content n order to reduce the total transfer te. Other work aed on dstrbuton large fles uses custo overlay protocols n peer-to-peer based fashon. Bt- Torrent [7] s the ost heavly deployed: t pleents peer-to-peer network wth centralzed trackng of avalable clents. However, ths syste has a low reslence to node falures, and could be sgnfcantly proved n ths drecton. Aong acadec ntatves are Bullet [] whch uses a self-organzng overlay esh, and SpltStrea [5] whch uses a collecton of superposed ultcast trees reseblng ALM-FastReplca constructon. SpltStrea supports the streang eda ode, and whle tryng to optze the streang qualty of delvered content, t does not attept to nze the download te. The goal s qute dfferent: t attepts to support the streang of lve eda n peer-topeer fashon. Whle CDNs were orgnally ntended for statc web content, they have been appled for delvery of streang e-

4 da as well. Delverng streang eda over the Internet s challengng due to a nuber of factors such as hgh bt rates, delay and loss senstvty. Most of the current work n ths drecton concentrates on how to prove the eda delvery fro the edge servers (or rror servers) to the end clents. In order to prove streang eda qualty, the latest work n ths drecton [, 3] proposes streang vdeo fro ultple edge servers (or rror stes), and n partcular, by cobnng the benefts of ultple descrpton codng (MDC) wth Internet path dversty. MDC codes a eda strea nto ultple copleentary descrptons. These descrptons have the property that f ether descrpton s receved t can be used to decode the baselne qualty vdeo, and ultple descrptons can be used to decode proved qualty vdeo. One of the basc assuptons n the research papers referred to above s that the orgnal content s already replcated across the edge (rror) servers. The goal of our paper s to address the content dstrbuton wthn ths nfrastructure (and not to the clents of ths nfrastructure). In our work, we propose a ethod to effcently replcate the content (represented by large fles) fro a sngle source to a large nuber of servers n a scalable and relable way. We explot deas of parttonng the orgnal fle and usng dverse Internet paths between the recpent nodes to speedup the dstrbuton of an orgnal large fle over Internet. 3 ALM-FastReplca Let N 0 be a node whch has an orgnal fle F. and letsze(f ) denote the sze of fle n bytes. The proble conssts n replcatng fle F across nodes N,..., N n whle nzng the overall (both average and axu) replcaton te. Orgnal FastReplca algorth [6] works as follows. Let k be a nuber of network connectons chosen for concurrent transfers between a sngle node and ultple recevng nodes. The orgnal set of nodes s parttoned nto replcaton groups, each consstng of k nodes. Fle s dvded n k equal subsequent subfles: F,..., F k. The replcaton process fro N 0 to a frst replcaton group s called FastReplca n the Sall, and t proceeds as follows. Step : Dstrbuton Step. The orgnator node N 0 opens k concurrent network connectons to nodes N,..., of the frst replcaton group, and sends to each recpent node N subfle ( k). Step : Collecton Step. After recevng fle, node N opens k concurrent network connectons to reanng nodes n the group and sends subfle to the (.e. a fle transfer s done va a store-and-forward ode). At ths step, each node N has the followng set of connectons: there are k outgong connectons fro node N : one connecton to each node N ( ) for sendng the correspondng subfle to node N. there are k ncong connectons to node N : one connecton fro each node N ( ) for sendng the correspondng subfle F to node N. At the end of the collecton step, each node receves all subfles F,..., F k coprsng the entre orgnal fle F. After that FastReplca n the sall s appled teratvely, usng nodes N,..., as the orgn nodes to replcate fle F further to new groups of nodes. Exaple. Let k = 0. Inthree algorth teratons (0 0 0), the orgnal fle can be replcated aong 000 nodes. At each teraton, the replcaton process follows FastReplca n the sall,.e. the teraton conssts of steps, each used for transferrng the k -th porton of the orgnal fle F. In ths work, we propose to cobne FastReplca schea wth applcaton level ultcast n the followng way. Let k be a nuber of network connectons chosen for concurrent transfers between a sngle node and ultple recevng nodes. The orgnal set of nodes s parttoned nto replcaton groups, each consstng of k nodes. Let G,..., G k be the correspondng replcaton groups. Let be a targeted nuber of groups coprsng a ultcast tree (accordngly to prevous studes [9], a reasonable value of ay vary n a range of several 0s). Then replcaton groups G,..., G k are arranged n the specal ultcast trees ˆM,M,..., M, each consstng of (or less) groups, where ˆM s called a prary ultcast tree, and M,..., M are called the secondary ultcast trees. To acheve the best perforance results, the values and should be slar: ths wll lead to well-balanced ultcast trees. Fle s dvded n k equal subsequent subfles: F,..., F k where Sze( )= Sze(F ) bytes for each k. k A hgh-level schea of ALM-FastReplca s shown n Fgure. Frst, ALM-FastReplca algorth replcates fle F va the prary ultcast tree ˆM. Once groups Ĝ,..., Ĝ, coprsng the prary ultcast tree ˆM, receve subfles F,..., F k copletely they ntate (ndependently fro each other) transfers of subfles F,..., F k to the secondary ultcast trees M,..., M. Now, we descrbe ALM-FastReplca n ore detal. ). Startng the prary ultcast tree ˆM. Let groups Ĝ,..., Ĝ coprse the prary ultcast tree ˆM. Let Ĝ = {N,..., Nk },. Dstrbuton step. Orgnator node N 0 opens k concurrent network connectons to nodes N,..., of Dependng on the nuber of nodes n the orgnal replcaton set, ALM-FastReplca ay operate by usng only a prary ultcast tree, or ay requre to eploy two-level schea as descrbed n the paper.

5 N 0 G (dstr) G (coll) G G G G Secondary ultcast trees G M Prary ultcast tree G M M Fgure : ALM-FastReplca: a hgh-level, overall schea. group Ĝ, and starts sendng subfle to the correspondng recpent node N ( k). Ths step s represented by box Ĝ(dstr) n Fgure. Collecton step. In group Ĝ, each node N after recevng frst packet of fle, edately starts sendng the fle to the rest of the nodes n group Ĝ (.e. a fle transfer s done va a fast-forward ode). In ths way, each node n group Ĝ wll be recevng all subfles F,..., F k of orgnal fle F. Ths step s represented by box Ĝ(coll) n Fgure. Group councaton step. Councaton aong groups Ĝ and Ĝ follows a dfferent fle exchange protocol defnng another typcal councaton pattern actvely used n the ALM-FastReplca algorth. We denote ths step as a group councaton step. Each node N of group Ĝ after recevng frst packet of fle, edately starts sendng the fle to node N of group Ĝ. After that nodes of group Ĝ perfor the collecton step descrbed above,.e. each node N opens k concurrent network connectons to the rest of the nodes of group Ĝ for transferrng ts subfle. In ths way, each node of group Ĝ wll be recevng all subfles F,..., F k of the orgnal fle F. The councatons aong the nodes n groups Ĝ and Ĝ are shown n ore detal n Fgure. N N N k G F F F k N N N G k F F F k G Slarly, group Ĝ starts councatons wth group Ĝ3 usng group councaton step edately after any node N receves the frst packet of the correspondng fle. Ths replcaton procedure contnues unrollng through the set of correspondng groups n ultcast tree ˆM shown n Fgure 3. In fact, as t s shown n Fgure 3, the ultcast tree ˆM s a collecton of k ultcast sub-trees ˆM F, ˆM F,..., ˆM Fk, where each such sub-tree ˆM F s replcatng the correspondng subfle. At the sae te, nodes fro these dfferent ultcast sub-trees use addtonal cross-connectons be- N N N G k Fgure : Group councaton step between the nodes n groups Ĝ and Ĝ. N F F F k N M F F N M F F k N k N N N k G F F k N N Nk G N N Nk F N F Prary Multcast Tree M N G - N 0 F k N N Multcast Sub-Trees F k k M Fk G G G Dstrbuton Step Collecton Step Group Councaton Step Group Councaton Step Fgure 3: Councatons aong the groups of prary ultcast tree ˆM. tween ther nodes (as shown n Fgure 3) to exchange ther copleentary subfles. ). Startng the secondary ultcast trees M,..., M. When node N of Ĝ receves the entre subfle F n the prary ultcast tree ˆM, then t starts transferrng fle F to group G of the secondary tree M usng the group councaton step. Thus, each group Ĝ ( ) of the prary ultcast tree ˆM ntates the replcaton process of subfles F,..., F k to the next, secondary ultcast tree M ={G,..., G } (see Fgure ). These transfers are asynchronous wthn the group Ĝ ={N,..., }. Ths copletes the descrpton of ALM-FastReplca.

6 4 ALM-FastReplca and MDC Multple Descrpton Codng (MDC) s source codng for ultple channels such that a decoder whch receves an arbtrary subset of the channels ay produce a useful reconstructon [8, 0, 5]. MDC codes a eda strea nto ultple copleentary descrptons. These descrptons have the property that f ether descrpton s receved t can be used to decode the baselne qualty vdeo, and ultple descrptons can be used to decode proved qualty vdeo. Ths s n contrast to conventonal vdeo coders (e.g. MPEG-//4, H.6/3, Mcrosofts and Real Network s propretary coders)., whch produce a sngle strea that does not have these MD propertes; we refer to these ethods as sngle descrpton codng (SDC). MDC refers to a for of copresson where a sgnal s coded nto a nuber of separated bt streas, where the ultple btstreas are referred to as ultple descrptons (MD). These ultple descrptons provde two portant propertes. Frst, each descrpton can be decoded ndependently to gve a usable reproducton of the orgnal sgnal. Second, the ultple descrptons contan copleentary nforaton so that the qualty of the decoded sgnal proved wth the nuber of descrptons that are correctly receved. An portant pont s that each descrpton or MD btstrea s ndependent of each other and s typcally of roughly equal portance. Layered or scalable approaches [7] essentally prortze data and thereby support ntellgent dscardng of the data (the enhanceent data can be lost or dscarded whle stll antanng usable vdeo). However the base-layer btstrea s crtcally portant f t s lost then the other btstrea(s) are useless. MD codng overcoes ths proble by allowng useful reproducton of the sgnal when any descrpton s receved, and wth ncreasng qualty when ore descrptons are receved. In order to prove streang eda qualty, the latest work n ths drecton proposes streang vdeo fro ultple edge servers (or rror stes), and n partcular, by cobnng the benefts of ultple descrpton codng (MDC) [] wth Internet path dversty. Under ths approach, the dfferent descrptons are explctly sent over dfferent routes to a clent. Path dversty explots the fact that whle any network lk ay suffer fro packet loss, there s a uch saller chance that two (or ore) network paths ay suffer sultaneously fro losses. In the orgnal ALM-FastReplca algorth, we partton the orgnal fle n n equal subsequent subfles and apply ALM-FastReplca to replcate the. Ths part of the algorth can be odfed accordngly to the nature of the fle. For exaple, for a eda fle encoded wth MDC, dfferent descrptons can be treated as subfles, and ALM- FastReplca can be appled to replcate the. Ths wll sgnfcantly prove the replcaton te. Takng nto account the nature of MDC (.e. that ether descrpton receved by the recpent node can be used to decode the baselne qualty vdeo), the edge servers n CDN ay start servng the correspondng eda fle once any of ts descrptons s receved. 5 Relable ALM-FastReplca In ths Secton, we extend the ALM-FastReplca algorth to be able to deal wth node falures. Basc algorth presented n Secton 3 s senstve to node falures. Under the fast-forward odel, a sngle node falure n a ultcast tree durng the fle dstrbuton ay pact the fle delvery to a sgnfcant nuber of nodes. For exaple, f node N (fro group Ĝ n ˆM) fals durng the ether dstrbuton or collecton steps descrbed n Secton 3 then ths event ay pact all nodes N,..., n the group Ĝ because each node depends on node N to replcate subfle F. A slar stuaton occurs f a falure of node N happens durng the group councaton step between groups Ĝ and Ĝ: ths falure ay pact all the nodes n the dependent subtree because the nodes n ths subtree should receve subfle fro node N. Thus, t s portant to desgn an algorth whch s able to deal wth node falures. It s a non-trval task because each node n the ultcast tree has a wde fan-out (0 or ore connectons out). Usng a nave approach where the tree below the faled node N faled s reconnected to ts predecessor N prev wll not be a well perforng soluton. Under such a soluton, N prev becoes a severe bottleneck n the whole syste snce t needs to handle the doubled nuber of connectons. Relable ALM-FastReplca proposed below effcently deals wth node falures by akng the local repar decson wthn the partcular group of nodes. It keeps the an structure of the ALM-FastReplca algorth practcally unchanged whle addng the desred property of reslence to node falures. Let us call group Ĝ an ntal group. We wll pay a specal attenton to group Ĝ and node falures n t, because t s the ost vulnerable and crucal group of nodes. Nodes of Ĝ are called ntal nodes. We wll call the rest of the groups subsequent groups, and ther nodes - subsequent nodes. There are several councaton patterns n whch node N ght be nvolved at oent of ts falure: If node N s an ntal node,.e. N = N, t ay fal: durng the dstrbuton step, when node N 0 s transferrng fle to N. Only node N 0 has subfle fle at ths pont. Snce node N s faled durng the fle transfer fro N 0 to N, node N 0 s aware of node N falure.

7 durng the collecton step, when node N has receved frst packet of fle and started to transfer t to the reanng nodes n group Ĝ. durng the group councaton step n the frst ultcast tree ˆM,.e. when node N s transferrng fle to the nodes of group Ĝ. durng the group councaton step to the next ultcast tree M,.e. when node N s transferrng fle to the nodes of group G as shown n Fgure. The crucal dfference of the node falure at ths step s that any node n group Ĝ already has receved subfle. If node N s a subsequent node, t ay fal durng the group councaton step when node N of group G s transferrng fle to the nodes of the next group G +. In the relable ALM-FastReplca algorth, node N of group G sends heart-beat essages to node N of group G. These heart-beat essages are augented wth addtonal nforaton on the correspondng algorth step and the current replcaton lst of nodes correspondng to ths step. Ths nforaton s necessary because of the asynchronous nature of the ALM-FastReplca algorth. For exaple, whle soe of the nodes of group Ĝ can perfor a fle transfer n the prary ultcast tree ˆM (.e. they are stll replcatng ther subfles to the correspondng nodes of group Ĝ), soe faster nodes of the sae group Ĝ ght already have started a fle transfer to the next ultcast tree M (.e. they are replcatng ther subfles to the correspondng nodes of group G as shown n Fgure ). Thus n case of node falure, t s portant to know: whch partcular node n the group has faled; whether the node s ntal or not; whether the correspondng step of the algorth s the dstrbuton, collecton,orgroup councaton step; whch ultcast tree, group and set of recevng nodes are pacted as a result of ths falure. There are dfferent repar procedures dependng on the crcustances under whch the node falure occurred: The ntal node N of group Ĝ fals durng dstrbuton, collecton, orgroup councaton step n the prary ultcast tree ˆM. In ths case, node N 0 s ether aware of the node N falure or receves a essage about ths fro the heartbeat group Ĝ. Snce node N 0 s the root of the overall replcaton procedure, to avod a sngle pont of falure t has a buddy-node ˆN 0 wth rrored nforaton and data. Node N 0 sends a essage to ˆN 0 to open k network connectons to the rest of the nodes n group Ĝ for sendng the ssng fle to each node n the group. Node N fals durng the group councaton step, where node N s ether a subsequent node, or an ntal node perforng the group councaton step to secondary ultcast tree. Fgure 4 depcts a falure of node N n group G whle t was transferrng fle to the rest of the nodes n group G and further to node N + of group G + at the group councaton step of the algorth. - N N - N N N N + N N + N - Faled Node N N F - G Faled Node G - G N + + G + Group Councaton Step Fgure 4: Node N falure durng group councaton step. The repar procedure for node N s shown n Fgure 5. Once node N realzed that node N has faled, node N conveys ths nforaton to the rest of the nodes n group G. The nodes n group G share the addtonal load of transferrng fle to the nodes of group G \ N as shown n Fgure 5. Node N transfers fle to node N + of group G +. - N N - - N - G - Faled Node N N N + N N N + N Faled Node N G G N + + G + Group Councato Step Fgure 5: Repar procedure for node N faled durng group councaton step. The dstrbuton lsts are the sae for all the nodes of group G, and after the repar step, the ALM- FastReplca algorth proceeds n the usual way for the entre subtree orgnated n group G. The nodes n group G contnue to share the addtonal load

8 of transferrng fle to the next, secondary ultcast tree orgnated at a group G or untl node N s repared. Snce the load of faled node N s shared aong k nodes of group G, the perforance degradaton s gradual for the repared porton of the dstrbuton tree. 6 Perforance Evaluaton Let Te (F ) denote the transfer te of fle F fro the orgnal node N 0 to node N as easured at node N.We use transfer te and replcaton te nterchangeably n the text. In our study, we consder the followng two perforance etrcs: Average replcaton te: Te aver = n n Te (F ) = Maxu replcaton te: Te ax = ax{te (F )}, {,,n} replcaton set receve a copy of the orgnal fle, and the prary goal of ALM-FastReplca s to nze the axu replcaton te. However, we are also nterested n understandng the pact of ALM-FastReplca on the average replcaton te Te aver. To deonstrate perforance benefts of ALM-FastReplca, let us consder a case wth 000 replcaton nodes, where each node can open k =0concurrent connectons,.e. there are 00 groups, each consstng of k =0nodes. Let Multple Uncast denote a schea that works teratvely n the followng way. Frst, t transfers the entre fle F fro the orgnal node to a frst group of 0 nodes by sultaneously openng k =0pont-to-pont concurrent network connectons. Then, t uses these 0 nodes (when they receve the entre fle F ) as the orgn nodes and repeats the sae procedure to the new groups of nodes, etc. The orgnal FastReplca algorth works teratvely too. Fle s dvded n k =0equal subsequent subfles and s replcated to a frst group of 0 nodes. Once they receve all the subfles, coprsng fle F, FastReplca uses these 0 nodes as the orgn nodes wth fle F and repeats the sae procedure agan, etc. Thus n three teratons, fle F can be replcated aong 000 nodes. Each teraton conssts of steps (dstrbuton and collecton), each used for transferrng the 0-th porton of the orgnal fle F. The ALM-FastReplca algorth operates as descrbed n Secton 3 usng a prary ultcast tree consstng of 0 replcaton groups, and 9 secondary ultcast trees, each also consstng of 0 replcaton groups. Let BW denote a bandwdth atrx, where BW[][] reflects the avalable bandwdth of the path fro node N to node N (as easured at soe te T ), and let Var be the rato of axu to nu avalable bandwdth along the paths partcpatng n the fle transfers. We call Var a bandwdth varaton. In our analyss, we consder the bandwdth atrx BW to be populated n the followng way: BW[][] =B rando(, Var ), where functon rando(, Var ) returns a rando nteger var: var Var. Whle t s a splstc odel, t helps to reflect a realstc stuaton, where the avalable bandwdth of dfferent lnks can be sgnfcantly dfferent. We wll call ths odel a unfor-rando odel. To perfor a senstvty analyss of how ALM- FastReplca perforance depends on a bandwdth varaton of partcpatng paths, we experented wth a range of dfferent values for Var between and 0. When Var =,t s an dealstc settng, where all of the paths are hoogeneous and have the sae bandwdth B (.e. no varaton n bandwdth). When Var =0, the network paths between the nodes have hghly varable avalable bandwdth wth a possble dfference of up to 0 tes. Usng the unfor-rando odel and ts bandwdth atrx BW, we coputed the average and axu fle replcaton tes under Multple Uncast, FastReplca, ALM- FastReplca for 000 nodes n the replcaton set. After that, we derved the relatve speedup of the fle replcaton te under ALM-FastReplca copared to the replcaton te under the Multple Uncast and FastReplca strateges. Relatve speedup of average fle replcaton te s shown n Fgure 6. Speedup (tes) Speedup: Average Replcaton Te 6 ALM-FR-aganst-MU 4 ALM-FR-aganst-FR Bandwdth Varaton Fgure 6: Unfor-rando odel: speedup n average fle replcaton te under ALM-FastReplca. For Var=, there s no varaton n bandwdth, and all the paths have the sae bandwdth B. Under ths dealstc settng, ALM-FastReplca outperfors FastReplca three

9 tes, and Multple Uncast ffteen tes as shown n Fgure 6. It can be explaned n the followng way. The ALM-FastReplca replcaton process conssts of two steps: ) prary ultcast tree and ) secondary ultcast trees, each used for transferrng the 0-th porton of the orgnal fle F. The FastReplca replcaton process conssts of three teratons. At each teraton, the replcaton process follows FastReplca n the sall,.e. the teraton conssts of steps, each used for transferrng the 0-th porton of the orgnal fle F. Thus overall FastReplca perfors 6 steps, each used for transferrng the 0-th porton of the orgnal fle F. Thus, the replcaton te under ALM-FastReplca s 3 tes better 6 than under FastReplca: 0 / 0 =3. Fnally, the Multple Uncast replcaton process conssts of three teratons, each used for transferrng the entre fle F. Thus, the replcaton te under ALM-FastReplca s 5 tes better than under Multple Uncast: 3 / 0 =5. Whle the perforance benefts of ALM-FastReplca are decreasng for hgher varaton of bandwdth of partcpatng paths, t stll reans very effcent, outperforng FastReplca ore than two tes, and Multple Uncast ore than fve tes n average replcaton te under bandwdth varaton of 0. Maxu replcaton te under ALM-FastReplca s three tes better than under FastReplca, and ffteen tes better than under Multple Uncast ndependent of the values of bandwdth varaton (ntutvely, these are proveents of the worst path n unfor-rando odel, and the coparson of replcaton te for a worst case scenaro under dfferent strateges s slar to dealstc settng case). 7 Concluson Streang eda over the Internet can be sgnfcantly proved wth specally desgned content dstrbuton technques and archtectures. In recent years, the Internet servces has oved fro an archtecture where data obects are located at a sngle orgn server to the archtecture where obects are replcated across ultple, geographcally dstrbuted servers. Clent requests for content are redrected to a best-suted replca rather than the orgn server. For large eda fles, the replcaton process across ths dstrbuted network of servers s a challengng, resource-ntensve proble on ts own. In ths work, we propose a new algorth, called ALM- FastReplca, for an effcent and relable replcaton of eda fles n the Internet envronent. Ths algorth cobnes the orgnal FastReplca deas wth ALM approach for sgnfcantly reducng the replcaton te of the large fles wthn CDN. We augent ALM-FastReplca wth an effcent relablty echans, that can deal wth node falures by akng local repar decsons wthn a partcular replcaton group of nodes. Proposed relablty algorth s constructed n such a way, that n the repared porton of the dstrbuton tree, the perforance degradaton s gradual, snce the load of the faled node s shared aong the nodes of the precedng replcaton group. References [] J. Apostolopoulos, T. Wong, W. Tan, and S. Wee. On ultple descrpton streang wth content delvery networks. Proc. of IEEE Infoco, 00. [] J. Byers, M.Luby, M. Mtzenacher, A. Rege. A Dgtal Fountan approach to relable dstrbuton of bulk data. Proc. of ACM SIG- COMM, 998. [3] J. Byers, M. Luby, M. Mtzenacher. Accessng ultple rror stes n parallel: Usng Tornado codes to speedup downloads. Proc. of IEEE Infoco, 999. [4] J. Byers, J. Consdne, M. Mtzenacher, S. Rost. Infored content delvery across adaptve overlay networks. Proc. of ACM SIGCOMM, 00. [5] M. Castro, P. Druschel, A. Kerarrec, A. Nand, A. Rowstron, A. Sngh. SpltStrea: Hgh-Bandwdth Multcast n Cooperatve Envronents. Proc. of ACM SOSP, 003. [6] L. Cherkasova, J. Lee. FastReplca: Effcent Large Fle Dstrbuton wthn Content Delvery Networks. Proc. of the 4th USENIX Syposu on Internet Technologes, March, 00. [7] B. Cohen. BtTorrent, [8] Y. Chu, S. Rao, H. Zhang. A case for end syste ultcast. Proc. of ACM SIGMETRICS, June, 000. [9] Y. Chu, S. Rao, S. Seshan, H. Zhang. Enablng conferencng applcatons on the Internet usng an overlay ultcast archtecture. Proc. of ACM SIGCOMM, 00. [0] V. Goyal, J. Kovacevc. Multple descrpton transfor codng of ages. Bell Labs, 998. [] J. Jannott, D. Gfford, K. Johnson, M. Kaashoek, J. O Toole, Jr. Overcast: Relable Multcastng wth an Overlay Network. Proc. of the 4th Syp. on Operatng Syste Desgn and Ipleentaton (OSDI), 000. [] D. Kostc, A. Rodrguez, J. Albrecht, A. Vahdat. Bullet: hgh bandwdth data dssenaton usng an overlay esh. Proc. of the 9th ACM SOSP 003. [3] T. Nguyen and A. Zakhor. Dstrbuted Vdeo Streang over the Internet. Proc. of SPIE Conference on Multeda Coputng and Networkng, San Jose, CA, 00. [4] D. Pendaraks, S. Sh, D. Vera, and M. Waldvogel. ALMI: an applcaton level ultcast nfrastructure. Proc. of USITS 00. [5] Justn Rdge, Fred W. Ware and Jerry D. Gbson. Multple Descrptons, Error Concealent, and Refned Descrptons for Iage Codng. Proc. Second Annual UCSD Conference on Wreless Councatons, 999. [6] P. Rodrguez, A. Krpal, and E. W. Bersack. Parallel-access for rror stes n the Internet. Proc. of IEEE Infoco, 000. [7] W. Tan and A. Zakhor. Real-Te nternet Vdeo usng Error-reslent Scalable Copresson and TCP-Frendly Transport Protocol. IEEE Trans. on Multeda, June, 999. [8] S. Wenger. Vdeo Redundancy Codng n H.63+. Workshop on Audo-Vsual Servces for packet networks, 997.

Multicast Tree Rearrangement to Recover Node Failures. in Overlay Multicast Networks

Multicast Tree Rearrangement to Recover Node Failures. in Overlay Multicast Networks Multcast Tree Rearrangeent to Recover Node Falures n Overlay Multcast Networks Hee K. Cho and Chae Y. Lee Dept. of Industral Engneerng, KAIST, 373-1 Kusung Dong, Taejon, Korea Abstract Overlay ultcast

More information

On-line Scheduling Algorithm with Precedence Constraint in Embeded Real-time System

On-line Scheduling Algorithm with Precedence Constraint in Embeded Real-time System 00 rd Internatonal Conference on Coputer and Electrcal Engneerng (ICCEE 00 IPCSIT vol (0 (0 IACSIT Press, Sngapore DOI: 077/IPCSIT0VNo80 On-lne Schedulng Algorth wth Precedence Constrant n Ebeded Real-te

More information

Introduction. Leslie Lamports Time, Clocks & the Ordering of Events in a Distributed System. Overview. Introduction Concepts: Time

Introduction. Leslie Lamports Time, Clocks & the Ordering of Events in a Distributed System. Overview. Introduction Concepts: Time Lesle Laports e, locks & the Orderng of Events n a Dstrbuted Syste Joseph Sprng Departent of oputer Scence Dstrbuted Systes and Securty Overvew Introducton he artal Orderng Logcal locks Orderng the Events

More information

Simulation Based Analysis of FAST TCP using OMNET++

Simulation Based Analysis of FAST TCP using OMNET++ Smulaton Based Analyss of FAST TCP usng OMNET++ Umar ul Hassan 04030038@lums.edu.pk Md Term Report CS678 Topcs n Internet Research Sprng, 2006 Introducton Internet traffc s doublng roughly every 3 months

More information

A Fast Content-Based Multimedia Retrieval Technique Using Compressed Data

A Fast Content-Based Multimedia Retrieval Technique Using Compressed Data A Fast Content-Based Multmeda Retreval Technque Usng Compressed Data Borko Furht and Pornvt Saksobhavvat NSF Multmeda Laboratory Florda Atlantc Unversty, Boca Raton, Florda 3343 ABSTRACT In ths paper,

More information

An Efficient Fault-Tolerant Multi-Bus Data Scheduling Algorithm Based on Replication and Deallocation

An Efficient Fault-Tolerant Multi-Bus Data Scheduling Algorithm Based on Replication and Deallocation BULGARIAN ACADEMY OF SCIENCES CYBERNETICS AND INFORMATION TECHNOLOGIES Volue 16, No Sofa 016 Prnt ISSN: 1311-970; Onlne ISSN: 1314-4081 DOI: 10.1515/cat-016-001 An Effcent Fault-Tolerant Mult-Bus Data

More information

A Unified Approach to Survivability of Connection-Oriented Networks

A Unified Approach to Survivability of Connection-Oriented Networks A Unfed Approach to Survvablty of Connecton-Orented Networs Krzysztof Walowa Char of Systes and Coputer Networs, Faculty of Electroncs, Wroclaw Unversty of Technology, Wybrzeze Wyspansego 27, 50-370 Wroclaw,

More information

Assignment # 2. Farrukh Jabeen Algorithms 510 Assignment #2 Due Date: June 15, 2009.

Assignment # 2. Farrukh Jabeen Algorithms 510 Assignment #2 Due Date: June 15, 2009. Farrukh Jabeen Algorthms 51 Assgnment #2 Due Date: June 15, 29. Assgnment # 2 Chapter 3 Dscrete Fourer Transforms Implement the FFT for the DFT. Descrbed n sectons 3.1 and 3.2. Delverables: 1. Concse descrpton

More information

Merging Results by Using Predicted Retrieval Effectiveness

Merging Results by Using Predicted Retrieval Effectiveness Mergng Results by Usng Predcted Retreval Effectveness Introducton Wen-Cheng Ln and Hsn-Hs Chen Departent of Coputer Scence and Inforaton Engneerng Natonal Tawan Unversty Tape, TAIWAN densln@nlg.cse.ntu.edu.tw;

More information

Video Proxy System for a Large-scale VOD System (DINA)

Video Proxy System for a Large-scale VOD System (DINA) Vdeo Proxy System for a Large-scale VOD System (DINA) KWUN-CHUNG CHAN #, KWOK-WAI CHEUNG *# #Department of Informaton Engneerng *Centre of Innovaton and Technology The Chnese Unversty of Hong Kong SHATIN,

More information

Estimating Progress of Execution for SQL Queries

Estimating Progress of Execution for SQL Queries Estatng Progress of Executon for SQL Queres Surajt Chaudhur Vvek arasayya Ravshankar Raaurthy Mcrosoft Research Mcrosoft Research Unversty of Wsconsn, Madson surajtc@crosoft.co vveknar@crosoft.co rav@cs.wsc.edu

More information

Efficient Distributed File System (EDFS)

Efficient Distributed File System (EDFS) Effcent Dstrbuted Fle System (EDFS) (Sem-Centralzed) Debessay(Debsh) Fesehaye, Rahul Malk & Klara Naherstedt Unversty of Illnos-Urbana Champagn Contents Problem Statement, Related Work, EDFS Desgn Rate

More information

Optimization Methods: Integer Programming Integer Linear Programming 1. Module 7 Lecture Notes 1. Integer Linear Programming

Optimization Methods: Integer Programming Integer Linear Programming 1. Module 7 Lecture Notes 1. Integer Linear Programming Optzaton Methods: Integer Prograng Integer Lnear Prograng Module Lecture Notes Integer Lnear Prograng Introducton In all the prevous lectures n lnear prograng dscussed so far, the desgn varables consdered

More information

Generating Fuzzy Term Sets for Software Project Attributes using and Real Coded Genetic Algorithms

Generating Fuzzy Term Sets for Software Project Attributes using and Real Coded Genetic Algorithms Generatng Fuzzy Ter Sets for Software Proect Attrbutes usng Fuzzy C-Means C and Real Coded Genetc Algorths Al Idr, Ph.D., ENSIAS, Rabat Alan Abran, Ph.D., ETS, Montreal Azeddne Zah, FST, Fes Internatonal

More information

A New Scheduling Algorithm for Servers

A New Scheduling Algorithm for Servers A New Schedulng Algorth for Servers Nann Yao, Wenbn Yao, Shaobn Ca, and Jun N College of Coputer Scence and Technology, Harbn Engneerng Unversty, Harbn, Chna {yaonann, yaowenbn, cashaobn, nun}@hrbeu.edu.cn

More information

Scheduling Workflow Applications on the Heterogeneous Cloud Resources

Scheduling Workflow Applications on the Heterogeneous Cloud Resources Indan Journal of Scence and Technology, Vol 8(2, DOI: 0.7485/jst/205/v82/57984, June 205 ISSN (rnt : 0974-6846 ISSN (Onlne : 0974-5645 Schedulng Workflow Applcatons on the Heterogeneous Cloud Resources

More information

Routing Time-Constrained Traffic in Wireless Sensor Networks

Routing Time-Constrained Traffic in Wireless Sensor Networks Routng Te-Constraned Traffc n Wreless Sensor Networks Keal Akkaya and Mohaed Youns Departent of Coputer Scence and Electrcal Engneerng Unversty of Maryland, Baltore County Baltore, MD 2250 keal youns@cs.ubc.edu

More information

Equation Based Rate Control and Multiple Connections for Adaptive Video Streaming in Cellular Networks

Equation Based Rate Control and Multiple Connections for Adaptive Video Streaming in Cellular Networks Equaton Based Rate Control and Multple Connectons for Adaptve Vdeo Streang n Cellular Networks A. Mardanan Dehkord and V. Tabataba Vakl Abstract Rate control s an portant ssue for vdeo streang n cellular

More information

Improvement of Spatial Resolution Using BlockMatching Based Motion Estimation and Frame. Integration

Improvement of Spatial Resolution Using BlockMatching Based Motion Estimation and Frame. Integration Improvement of Spatal Resoluton Usng BlockMatchng Based Moton Estmaton and Frame Integraton Danya Suga and Takayuk Hamamoto Graduate School of Engneerng, Tokyo Unversty of Scence, 6-3-1, Nuku, Katsuska-ku,

More information

The Codesign Challenge

The Codesign Challenge ECE 4530 Codesgn Challenge Fall 2007 Hardware/Software Codesgn The Codesgn Challenge Objectves In the codesgn challenge, your task s to accelerate a gven software reference mplementaton as fast as possble.

More information

An Adaptive Sleep Strategy for Energy Conservation in Wireless Sensor Networks

An Adaptive Sleep Strategy for Energy Conservation in Wireless Sensor Networks An Adaptve Sleep Strategy for Energy Conservaton n Wreless Sensor Networks Guseppe Anastas, Marco Cont, Maro D Francesco Abstract - In recent years, wreless sensor network deployents for real lfe applcatons

More information

A study of Adaptive Replication Technique in routing time-constrained messages (VoIP) in MANET

A study of Adaptive Replication Technique in routing time-constrained messages (VoIP) in MANET A study of Adaptve Replcaton Technque n routng te-constraned essages (VoIP) n MANET SWATI SAXENA Research Scholar AIM & ACT Banasthal Vdyapth Banasthal, (Ra.) Dr. MADHAVI SINHA Asst. Prof. (HOD) Dept.

More information

Energy-aware Delay-Constrained Routing in Wireless Sensor Networks

Energy-aware Delay-Constrained Routing in Wireless Sensor Networks Energy-aware Delay-Constraned Routng n Wreless Sensor Networks Keal Akkaya and Mohaed Youns Departent of Coputer Scence and Electrcal Engneerng Unversty of Maryland Baltore County, Baltore, MD 21250 {keal1,

More information

ARTICLE IN PRESS. Signal Processing: Image Communication

ARTICLE IN PRESS. Signal Processing: Image Communication Sgnal Processng: Image Communcaton 23 (2008) 754 768 Contents lsts avalable at ScenceDrect Sgnal Processng: Image Communcaton journal homepage: www.elsever.com/locate/mage Dstrbuted meda rate allocaton

More information

Relevance Feedback in Content-based 3D Object Retrieval A Comparative Study

Relevance Feedback in Content-based 3D Object Retrieval A Comparative Study 753 Coputer-Aded Desgn and Applcatons 008 CAD Solutons, LLC http://www.cadanda.co Relevance Feedback n Content-based 3D Object Retreval A Coparatve Study Panagots Papadaks,, Ioanns Pratkaks, Theodore Trafals

More information

Color Image Segmentation Based on Adaptive Local Thresholds

Color Image Segmentation Based on Adaptive Local Thresholds Color Iage Segentaton Based on Adaptve Local Thresholds ETY NAVON, OFE MILLE *, AMI AVEBUCH School of Coputer Scence Tel-Avv Unversty, Tel-Avv, 69978, Israel E-Mal * : llero@post.tau.ac.l Fax nuber: 97-3-916084

More information

Solutions to Programming Assignment Five Interpolation and Numerical Differentiation

Solutions to Programming Assignment Five Interpolation and Numerical Differentiation College of Engneerng and Coputer Scence Mechancal Engneerng Departent Mechancal Engneerng 309 Nuercal Analyss of Engneerng Systes Sprng 04 Nuber: 537 Instructor: Larry Caretto Solutons to Prograng Assgnent

More information

Parallel matrix-vector multiplication

Parallel matrix-vector multiplication Appendx A Parallel matrx-vector multplcaton The reduced transton matrx of the three-dmensonal cage model for gel electrophoress, descrbed n secton 3.2, becomes excessvely large for polymer lengths more

More information

A Binarization Algorithm specialized on Document Images and Photos

A Binarization Algorithm specialized on Document Images and Photos A Bnarzaton Algorthm specalzed on Document mages and Photos Ergna Kavalleratou Dept. of nformaton and Communcaton Systems Engneerng Unversty of the Aegean kavalleratou@aegean.gr Abstract n ths paper, a

More information

Scalable Download Protocols

Scalable Download Protocols Scalable Download Protocols A Thess Submtted to the College of Graduate Studes and Research In Partal Fulfllment of the Requrements For the Degree of Doctor of Phlosophy In the Department of Computer Scence

More information

SRB: Shared Running Buffers in Proxy to Exploit Memory Locality of Multiple Streaming Media Sessions

SRB: Shared Running Buffers in Proxy to Exploit Memory Locality of Multiple Streaming Media Sessions SRB: Shared Runnng Buffers n Proxy to Explot Memory Localty of Multple Streamng Meda Sessons Songqng Chen,BoShen, Yong Yan, Sujoy Basu, and Xaodong Zhang Department of Computer Scence Moble and Meda System

More information

Load Balancing Algorithm for Web Server Based on Weighted Minimal Connections

Load Balancing Algorithm for Web Server Based on Weighted Minimal Connections Journal of Web ystes and Applcatons (2017) Vol. 1, Nuuber 1 Clausus centfc Press, Canada Load Balancng Algorth for Web erver Based on Weghted Mnal Connectons Pan Zhu1, a, Jangxng Zhang 2,b College of Inforaton

More information

Private Information Retrieval (PIR)

Private Information Retrieval (PIR) 2 Levente Buttyán Problem formulaton Alce wants to obtan nformaton from a database, but she does not want the database to learn whch nformaton she wanted e.g., Alce s an nvestor queryng a stock-market

More information

An Optimal Algorithm for Prufer Codes *

An Optimal Algorithm for Prufer Codes * J. Software Engneerng & Applcatons, 2009, 2: 111-115 do:10.4236/jsea.2009.22016 Publshed Onlne July 2009 (www.scrp.org/journal/jsea) An Optmal Algorthm for Prufer Codes * Xaodong Wang 1, 2, Le Wang 3,

More information

Parallelism for Nested Loops with Non-uniform and Flow Dependences

Parallelism for Nested Loops with Non-uniform and Flow Dependences Parallelsm for Nested Loops wth Non-unform and Flow Dependences Sam-Jn Jeong Dept. of Informaton & Communcaton Engneerng, Cheonan Unversty, 5, Anseo-dong, Cheonan, Chungnam, 330-80, Korea. seong@cheonan.ac.kr

More information

Handover Delay Estimation Method for IMS-Based Systems with Heterogeneous Access Networks

Handover Delay Estimation Method for IMS-Based Systems with Heterogeneous Access Networks Councatons 07; (): -7 http://www.scencepublshnggroup.co/j/co do: 0.8/j.co.0700. ISSN: 8-9 (Prnt); ISSN: 8-9 (Onlne) Handover Delay Estaton Method for IMS-Based Systes wth Heterogeneous Access Networks

More information

Content Classification for Caching under CDNs

Content Classification for Caching under CDNs Content Classfcaton for Cachng under CDNs George Palls 1,2, Charlaos homos 2, Konstantnos Stamos 2, Athena Vakal 2, George Andreads 3 1 Department of Computer Scence, Unversty of Cyprus 2 Department of

More information

126 Int. J. Ad Hoc and Ubiquitous Computing, Vol. 1, No. 3, Djamel Djenouri*

126 Int. J. Ad Hoc and Ubiquitous Computing, Vol. 1, No. 3, Djamel Djenouri* 126 Int. J. Ad Hoc and Ubqutous Coputng, Vol. 1, No. 3, 2006 New power-aware routng protocol for oble ad hoc networks Djael Djenour* Basc Software Laboratory, CERIST Center of Research, Algers, Algera

More information

Wishing you all a Total Quality New Year!

Wishing you all a Total Quality New Year! Total Qualty Management and Sx Sgma Post Graduate Program 214-15 Sesson 4 Vnay Kumar Kalakband Assstant Professor Operatons & Systems Area 1 Wshng you all a Total Qualty New Year! Hope you acheve Sx sgma

More information

Avoiding congestion through dynamic load control

Avoiding congestion through dynamic load control Avodng congeston through dynamc load control Vasl Hnatyshn, Adarshpal S. Seth Department of Computer and Informaton Scences, Unversty of Delaware, Newark, DE 976 ABSTRACT The current best effort approach

More information

Load Balancing for Hex-Cell Interconnection Network

Load Balancing for Hex-Cell Interconnection Network Int. J. Communcatons, Network and System Scences,,, - Publshed Onlne Aprl n ScRes. http://www.scrp.org/journal/jcns http://dx.do.org/./jcns.. Load Balancng for Hex-Cell Interconnecton Network Saher Manaseer,

More information

Predicting Power Grid Component Outage In Response to Extreme Events. S. BAHRAMIRAD ComEd USA

Predicting Power Grid Component Outage In Response to Extreme Events. S. BAHRAMIRAD ComEd USA 1, rue d Artos, F-75008 PARIS CIGRE US Natonal Cottee http : //www.cgre.org 016 Grd of the Future Syposu Predctng Power Grd Coponent Outage In Response to Extree Events R. ESKANDARPOUR, A. KHODAEI Unversty

More information

Related-Mode Attacks on CTR Encryption Mode

Related-Mode Attacks on CTR Encryption Mode Internatonal Journal of Network Securty, Vol.4, No.3, PP.282 287, May 2007 282 Related-Mode Attacks on CTR Encrypton Mode Dayn Wang, Dongda Ln, and Wenlng Wu (Correspondng author: Dayn Wang) Key Laboratory

More information

kccvoip.com basic voip training NAT/PAT extract 2008

kccvoip.com basic voip training NAT/PAT extract 2008 kccvop.com basc vop tranng NAT/PAT extract 28 As we have seen n the prevous sldes, SIP and H2 both use addressng nsde ther packets to rely nformaton. Thnk of an envelope where we place the addresses of

More information

Performance Analysis of Coiflet Wavelet and Moment Invariant Feature Extraction for CT Image Classification using SVM

Performance Analysis of Coiflet Wavelet and Moment Invariant Feature Extraction for CT Image Classification using SVM Perforance Analyss of Coflet Wavelet and Moent Invarant Feature Extracton for CT Iage Classfcaton usng SVM N. T. Renukadev, Assstant Professor, Dept. of CT-UG, Kongu Engneerng College, Perundura Dr. P.

More information

A MOVING MESH APPROACH FOR SIMULATION BUDGET ALLOCATION ON CONTINUOUS DOMAINS

A MOVING MESH APPROACH FOR SIMULATION BUDGET ALLOCATION ON CONTINUOUS DOMAINS Proceedngs of the Wnter Smulaton Conference M E Kuhl, N M Steger, F B Armstrong, and J A Jones, eds A MOVING MESH APPROACH FOR SIMULATION BUDGET ALLOCATION ON CONTINUOUS DOMAINS Mark W Brantley Chun-Hung

More information

2016 International Conference on Sustainable Energy, Environment and Information Engineering (SEEIE 2016) ISBN:

2016 International Conference on Sustainable Energy, Environment and Information Engineering (SEEIE 2016) ISBN: 06 Internatonal Conference on Sustanable Energy, Envronent and Inforaton Engneerng (SEEIE 06) ISBN: 978--60595-337-3 A Study on IEEE 80. MAC Layer Msbehavor under Dfferent Back-off Algorths Trong Mnh HOANG,,

More information

Cluster Analysis of Electrical Behavior

Cluster Analysis of Electrical Behavior Journal of Computer and Communcatons, 205, 3, 88-93 Publshed Onlne May 205 n ScRes. http://www.scrp.org/ournal/cc http://dx.do.org/0.4236/cc.205.350 Cluster Analyss of Electrcal Behavor Ln Lu Ln Lu, School

More information

Key-Words: - Under sear Hydrothermal vent image; grey; blue chroma; OTSU; FCM

Key-Words: - Under sear Hydrothermal vent image; grey; blue chroma; OTSU; FCM A Fast and Effectve Segentaton Algorth for Undersea Hydrotheral Vent Iage FUYUAN PENG 1 QIAN XIA 1 GUOHUA XU 2 XI YU 1 LIN LUO 1 Electronc Inforaton Engneerng Departent of Huazhong Unversty of Scence and

More information

Machine Learning: Algorithms and Applications

Machine Learning: Algorithms and Applications 14/05/1 Machne Learnng: Algorthms and Applcatons Florano Zn Free Unversty of Bozen-Bolzano Faculty of Computer Scence Academc Year 011-01 Lecture 10: 14 May 01 Unsupervsed Learnng cont Sldes courtesy of

More information

AN ALGORITHM FOR ODD GRACEFULNESS OF THE TENSOR PRODUCT OF TWO LINE GRAPHS

AN ALGORITHM FOR ODD GRACEFULNESS OF THE TENSOR PRODUCT OF TWO LINE GRAPHS Internatonal ournal on applcatons of graph theory n wreless ad hoc networks and sensor networks (GRAPH-HOC) Vol.3, No., March 0 AN ALGORITHM FOR ODD GRACEFULNESS OF THE TENSOR PRODUCT OF TWO LINE GRAPHS

More information

Advanced Computer Networks

Advanced Computer Networks Char of Network Archtectures and Servces Department of Informatcs Techncal Unversty of Munch Note: Durng the attendance check a stcker contanng a unque QR code wll be put on ths exam. Ths QR code contans

More information

Array transposition in CUDA shared memory

Array transposition in CUDA shared memory Array transposton n CUDA shared memory Mke Gles February 19, 2014 Abstract Ths short note s nspred by some code wrtten by Jeremy Appleyard for the transposton of data through shared memory. I had some

More information

DEAR: A DEVICE AND ENERGY AWARE ROUTING PROTOCOL FOR MOBILE AD HOC NETWORKS

DEAR: A DEVICE AND ENERGY AWARE ROUTING PROTOCOL FOR MOBILE AD HOC NETWORKS DEAR: A DEVICE AND ENERGY AWARE ROUTING PROTOCOL FOR MOBILE AD HOC NETWORKS Arun Avudanayagam Yuguang Fang Wenjng Lou Department of Electrcal and Computer Engneerng Unversty of Florda Ganesvlle, FL 3261

More information

Shared Running Buffer Based Proxy Caching of Streaming Sessions

Shared Running Buffer Based Proxy Caching of Streaming Sessions Shared Runnng Buffer Based Proxy Cachng of Streamng Sessons Songqng Chen, Bo Shen, Yong Yan, Sujoy Basu Moble and Meda Systems Laboratory HP Laboratores Palo Alto HPL-23-47 March th, 23* E-mal: sqchen@cs.wm.edu,

More information

Internet Traffic Managers

Internet Traffic Managers Internet Traffc Managers Ibrahm Matta matta@cs.bu.edu www.cs.bu.edu/faculty/matta Computer Scence Department Boston Unversty Boston, MA 225 Jont work wth members of the WING group: Azer Bestavros, John

More information

A New Token Allocation Algorithm for TCP Traffic in Diffserv Network

A New Token Allocation Algorithm for TCP Traffic in Diffserv Network A New Token Allocaton Algorthm for TCP Traffc n Dffserv Network A New Token Allocaton Algorthm for TCP Traffc n Dffserv Network S. Sudha and N. Ammasagounden Natonal Insttute of Technology, Truchrappall,

More information

A fair buffer allocation scheme

A fair buffer allocation scheme A far buffer allocaton scheme Juha Henanen and Kalev Klkk Telecom Fnland P.O. Box 228, SF-330 Tampere, Fnland E-mal: juha.henanen@tele.f Abstract An approprate servce for data traffc n ATM networks requres

More information

Performance Evaluation of Information Retrieval Systems

Performance Evaluation of Information Retrieval Systems Why System Evaluaton? Performance Evaluaton of Informaton Retreval Systems Many sldes n ths secton are adapted from Prof. Joydeep Ghosh (UT ECE) who n turn adapted them from Prof. Dk Lee (Unv. of Scence

More information

Real-time interactive applications

Real-time interactive applications Real-tme nteractve applcatons PC-2-PC phone PC-2-phone Dalpad Net2phone vdeoconference Webcams Now we look at a PC-2-PC Internet phone example n detal Internet phone over best-effort (1) Best effort packet

More information

CACHE MEMORY DESIGN FOR INTERNET PROCESSORS

CACHE MEMORY DESIGN FOR INTERNET PROCESSORS CACHE MEMORY DESIGN FOR INTERNET PROCESSORS WE EVALUATE A SERIES OF THREE PROGRESSIVELY MORE AGGRESSIVE ROUTING-TABLE CACHE DESIGNS AND DEMONSTRATE THAT THE INCORPORATION OF HARDWARE CACHES INTO INTERNET

More information

A mathematical programming approach to the analysis, design and scheduling of offshore oilfields

A mathematical programming approach to the analysis, design and scheduling of offshore oilfields 17 th European Symposum on Computer Aded Process Engneerng ESCAPE17 V. Plesu and P.S. Agach (Edtors) 2007 Elsever B.V. All rghts reserved. 1 A mathematcal programmng approach to the analyss, desgn and

More information

A system based on a modified version of the FCM algorithm for profiling Web users from access log

A system based on a modified version of the FCM algorithm for profiling Web users from access log A syste based on a odfed verson of the FCM algorth for proflng Web users fro access log Paolo Corsn, Laura De Dosso, Beatrce Lazzern, Francesco Marcellon Dpartento d Ingegnera dell Inforazone va Dotsalv,

More information

Pattern Classification of Back-Propagation Algorithm Using Exclusive Connecting Network

Pattern Classification of Back-Propagation Algorithm Using Exclusive Connecting Network World Acade of Scence, Engneerng and Technolog 36 7 Pattern Classfcaton of Bac-Propagaton Algorth Usng Eclusve Connectng Networ Insung Jung, and G-Na Wang Abstract The obectve of ths paper s to a desgn

More information

Sample Solution. Advanced Computer Networks P 1 P 2 P 3 P 4 P 5. Module: IN2097 Date: Examiner: Prof. Dr.-Ing. Georg Carle Exam: Final exam

Sample Solution. Advanced Computer Networks P 1 P 2 P 3 P 4 P 5. Module: IN2097 Date: Examiner: Prof. Dr.-Ing. Georg Carle Exam: Final exam Char of Network Archtectures and Servces Department of Informatcs Techncal Unversty of Munch Note: Durng the attendance check a stcker contanng a unque QR code wll be put on ths exam. Ths QR code contans

More information

The ray density estimation of a CT system by a supervised learning algorithm

The ray density estimation of a CT system by a supervised learning algorithm Te ray densty estaton of a CT syste by a suervsed learnng algort Nae : Jongduk Baek Student ID : 5459 Toc y toc s to fnd te ray densty of a new CT syste by usng te learnng algort Background Snce te develoent

More information

Multiple Instance Learning via Multiple Kernel Learning *

Multiple Instance Learning via Multiple Kernel Learning * The Nnth nternatonal Syposu on Operatons Research and ts Applcatons (SORA 10) Chengdu-Juzhagou, Chna, August 19 23, 2010 Copyrght 2010 ORSC & APORC, pp. 160 167 ultple nstance Learnng va ultple Kernel

More information

Multi-Constraint Multi-Processor Resource Allocation

Multi-Constraint Multi-Processor Resource Allocation Mult-Constrant Mult-Processor Resource Allocaton Ar R. B. Behrouzan 1, Dp Goswa 1, Twan Basten 1,2, Marc Gelen 1, Had Alzadeh Ara 1 1 Endhoven Unversty of Technology, Endhoven, The Netherlands 2 TNO Ebedded

More information

RAP. Speed/RAP/CODA. Real-time Systems. Modeling the sensor networks. Real-time Systems. Modeling the sensor networks. Real-time systems:

RAP. Speed/RAP/CODA. Real-time Systems. Modeling the sensor networks. Real-time Systems. Modeling the sensor networks. Real-time systems: Speed/RAP/CODA Presented by Octav Chpara Real-tme Systems Many wreless sensor network applcatons requre real-tme support Survellance and trackng Border patrol Fre fghtng Real-tme systems: Hard real-tme:

More information

TN348: Openlab Module - Colocalization

TN348: Openlab Module - Colocalization TN348: Openlab Module - Colocalzaton Topc The Colocalzaton module provdes the faclty to vsualze and quantfy colocalzaton between pars of mages. The Colocalzaton wndow contans a prevew of the two mages

More information

Scalable Video Streaming over P2P Networks: A Matter of Harmony?

Scalable Video Streaming over P2P Networks: A Matter of Harmony? 211 IEEE 16th Internatonal Workshop on Computer Aded Modelng and Desgn of Communcaton Lnks and Networks (CAMAD) Scalable Vdeo Streamng over P2P Networks: A Matter of Harmony? Samr Medah 1, Toufk Ahmed

More information

Concurrent Apriori Data Mining Algorithms

Concurrent Apriori Data Mining Algorithms Concurrent Apror Data Mnng Algorthms Vassl Halatchev Department of Electrcal Engneerng and Computer Scence York Unversty, Toronto October 8, 2015 Outlne Why t s mportant Introducton to Assocaton Rule Mnng

More information

Distortion-Memory Tradeoffs in Cache-Aided Wireless Video Delivery

Distortion-Memory Tradeoffs in Cache-Aided Wireless Video Delivery Dstorton-Memory Tradeoffs n Cache-Aded Wreless Vdeo Delvery P. Hassanzadeh, E. Erkp, J. Llorca, A. Tulno arxv:1511.03932v1 [cs.it] 12 Nov 2015 Abstract Moble network operators are consderng cachng as one

More information

Problem Set 3 Solutions

Problem Set 3 Solutions Introducton to Algorthms October 4, 2002 Massachusetts Insttute of Technology 6046J/18410J Professors Erk Demane and Shaf Goldwasser Handout 14 Problem Set 3 Solutons (Exercses were not to be turned n,

More information

Improving Low Density Parity Check Codes Over the Erasure Channel. The Nelder Mead Downhill Simplex Method. Scott Stransky

Improving Low Density Parity Check Codes Over the Erasure Channel. The Nelder Mead Downhill Simplex Method. Scott Stransky Improvng Low Densty Party Check Codes Over the Erasure Channel The Nelder Mead Downhll Smplex Method Scott Stransky Programmng n conjuncton wth: Bors Cukalovc 18.413 Fnal Project Sprng 2004 Page 1 Abstract

More information

A Semi-Distributed Load Balancing Architecture and Algorithm for Heterogeneous Wireless Networks

A Semi-Distributed Load Balancing Architecture and Algorithm for Heterogeneous Wireless Networks A Sem-Dstrbuted oad Balancng Archtecture and Algorthm for Heterogeneous reless Networks Md. Golam Rabul Ala Choong Seon Hong * Kyung Hee Unversty, Korea rob@networkng.khu.ac.kr, cshong@khu.ac.kr Abstract

More information

Term Weighting Classification System Using the Chi-square Statistic for the Classification Subtask at NTCIR-6 Patent Retrieval Task

Term Weighting Classification System Using the Chi-square Statistic for the Classification Subtask at NTCIR-6 Patent Retrieval Task Proceedngs of NTCIR-6 Workshop Meetng, May 15-18, 2007, Tokyo, Japan Term Weghtng Classfcaton System Usng the Ch-square Statstc for the Classfcaton Subtask at NTCIR-6 Patent Retreval Task Kotaro Hashmoto

More information

QoS-aware routing for heterogeneous layered unicast transmissions in wireless mesh networks with cooperative network coding

QoS-aware routing for heterogeneous layered unicast transmissions in wireless mesh networks with cooperative network coding Tarno et al. EURASIP Journal on Wreless Communcatons and Networkng 214, 214:81 http://wcn.euraspournals.com/content/214/1/81 RESEARCH Open Access QoS-aware routng for heterogeneous layered uncast transmssons

More information

Support Strong Consistency for Mobile Dynamic Contents Delivery Network

Support Strong Consistency for Mobile Dynamic Contents Delivery Network Nnth IEEE Internatonal Symposum on Multmeda 27 Support Strong Consstency for Moble Dynamc Contents Delvery Networ Zhou Su, Jro Katto and Yasuho Yasuda Graduate School of Scence and Engneerng, Waseda Unversty,

More information

Video streaming over the Internet is becoming very popular and

Video streaming over the Internet is becoming very popular and Streamng MPEG-4 AudoVsual Objects Usng TCP-Frendly Rate Control and Unequal Error Protecton Toufk Ahmed 1, Ahmed Mehaoua 1 and Vncent Lecure 2 1 2 CNRS-PRSM LabUnversty of Versalles CRAN lab CNRS UMR 739

More information

VRT012 User s guide V0.1. Address: Žirmūnų g. 27, Vilnius LT-09105, Phone: (370-5) , Fax: (370-5) ,

VRT012 User s guide V0.1. Address: Žirmūnų g. 27, Vilnius LT-09105, Phone: (370-5) , Fax: (370-5) , VRT012 User s gude V0.1 Thank you for purchasng our product. We hope ths user-frendly devce wll be helpful n realsng your deas and brngng comfort to your lfe. Please take few mnutes to read ths manual

More information

Some Advanced SPC Tools 1. Cumulative Sum Control (Cusum) Chart For the data shown in Table 9-1, the x chart can be generated.

Some Advanced SPC Tools 1. Cumulative Sum Control (Cusum) Chart For the data shown in Table 9-1, the x chart can be generated. Some Advanced SP Tools 1. umulatve Sum ontrol (usum) hart For the data shown n Table 9-1, the x chart can be generated. However, the shft taken place at sample #21 s not apparent. 92 For ths set samples,

More information

Routing in Degree-constrained FSO Mesh Networks

Routing in Degree-constrained FSO Mesh Networks Internatonal Journal of Hybrd Informaton Technology Vol., No., Aprl, 009 Routng n Degree-constraned FSO Mesh Networks Zpng Hu, Pramode Verma, and James Sluss Jr. School of Electrcal & Computer Engneerng

More information

End-to-end Distortion Estimation for RD-based Robust Delivery of Pre-compressed Video

End-to-end Distortion Estimation for RD-based Robust Delivery of Pre-compressed Video End-to-end Dstorton Estmaton for RD-based Robust Delvery of Pre-compressed Vdeo Ru Zhang, Shankar L. Regunathan and Kenneth Rose Department of Electrcal and Computer Engneerng Unversty of Calforna, Santa

More information

Supporting Consistency Management in Dynamic Content Distribution Overlays

Supporting Consistency Management in Dynamic Content Distribution Overlays Supportng Consstency Management n Dynamc Content Dstrbuton Overlays Zhou Su, Jro Katto, Yasuhko Yasuda Faculty of Scence and engneerng, Waseda Unversty Contact E-mal: suzhou@waseda.p Abstract Content Dstrbuton

More information

Module Management Tool in Software Development Organizations

Module Management Tool in Software Development Organizations Journal of Computer Scence (5): 8-, 7 ISSN 59-66 7 Scence Publcatons Management Tool n Software Development Organzatons Ahmad A. Al-Rababah and Mohammad A. Al-Rababah Faculty of IT, Al-Ahlyyah Amman Unversty,

More information

A Theory of Non-Deterministic Networks

A Theory of Non-Deterministic Networks A Theory of Non-Deternstc Networs Alan Mshcheno and Robert K rayton Departent of EECS, Unversty of Calforna at ereley {alan, brayton}@eecsbereleyedu Abstract oth non-deterns and ult-level networs copactly

More information

Load-Balanced Anycast Routing

Load-Balanced Anycast Routing Load-Balanced Anycast Routng Chng-Yu Ln, Jung-Hua Lo, and Sy-Yen Kuo Department of Electrcal Engneerng atonal Tawan Unversty, Tape, Tawan sykuo@cc.ee.ntu.edu.tw Abstract For fault-tolerance and load-balance

More information

AP PHYSICS B 2008 SCORING GUIDELINES

AP PHYSICS B 2008 SCORING GUIDELINES AP PHYSICS B 2008 SCORING GUIDELINES General Notes About 2008 AP Physcs Scorng Gudelnes 1. The solutons contan the most common method of solvng the free-response questons and the allocaton of ponts for

More information

A Semantic Model for Video Based Face Recognition

A Semantic Model for Video Based Face Recognition Proceedng of the IEEE Internatonal Conference on Inforaton and Autoaton Ynchuan, Chna, August 2013 A Seantc Model for Vdeo Based Face Recognton Dhong Gong, Ka Zhu, Zhfeng L, and Yu Qao Shenzhen Key Lab

More information

A Fast Visual Tracking Algorithm Based on Circle Pixels Matching

A Fast Visual Tracking Algorithm Based on Circle Pixels Matching A Fast Vsual Trackng Algorthm Based on Crcle Pxels Matchng Zhqang Hou hou_zhq@sohu.com Chongzhao Han czhan@mal.xjtu.edu.cn Ln Zheng Abstract: A fast vsual trackng algorthm based on crcle pxels matchng

More information

Optimally Combining Positive and Negative Features for Text Categorization

Optimally Combining Positive and Negative Features for Text Categorization Optally Cobnng Postve and Negatve Features for Text Categorzaton Zhaohu Zheng ZZHENG3@CEDAR.BUFFALO.EDU Rohn Srhar ROHINI@CEDAR.BUFFALO.EDU CEDAR, Dept. of Coputer Scence and Engneerng, State Unversty

More information

6.854 Advanced Algorithms Petar Maymounkov Problem Set 11 (November 23, 2005) With: Benjamin Rossman, Oren Weimann, and Pouya Kheradpour

6.854 Advanced Algorithms Petar Maymounkov Problem Set 11 (November 23, 2005) With: Benjamin Rossman, Oren Weimann, and Pouya Kheradpour 6.854 Advanced Algorthms Petar Maymounkov Problem Set 11 (November 23, 2005) Wth: Benjamn Rossman, Oren Wemann, and Pouya Kheradpour Problem 1. We reduce vertex cover to MAX-SAT wth weghts, such that the

More information

an assocated logc allows the proof of safety and lveness propertes. The Unty model nvolves on the one hand a programmng language and, on the other han

an assocated logc allows the proof of safety and lveness propertes. The Unty model nvolves on the one hand a programmng language and, on the other han UNITY as a Tool for Desgn and Valdaton of a Data Replcaton System Phlppe Quennec Gerard Padou CENA IRIT-ENSEEIHT y Nnth Internatonal Conference on Systems Engneerng Unversty of Nevada, Las Vegas { 14-16

More information

User Behavior Recognition based on Clustering for the Smart Home

User Behavior Recognition based on Clustering for the Smart Home 3rd WSEAS Internatonal Conference on REMOTE SENSING, Vence, Italy, Noveber 2-23, 2007 52 User Behavor Recognton based on Clusterng for the Sart Hoe WOOYONG CHUNG, JAEHUN LEE, SUKHYUN YUN, SOOHAN KIM* AND

More information

R s s f. m y s. SPH3UW Unit 7.3 Spherical Concave Mirrors Page 1 of 12. Notes

R s s f. m y s. SPH3UW Unit 7.3 Spherical Concave Mirrors Page 1 of 12. Notes SPH3UW Unt 7.3 Sphercal Concave Mrrors Page 1 of 1 Notes Physcs Tool box Concave Mrror If the reflectng surface takes place on the nner surface of the sphercal shape so that the centre of the mrror bulges

More information

Evaluation of an Enhanced Scheme for High-level Nested Network Mobility

Evaluation of an Enhanced Scheme for High-level Nested Network Mobility IJCSNS Internatonal Journal of Computer Scence and Network Securty, VOL.15 No.10, October 2015 1 Evaluaton of an Enhanced Scheme for Hgh-level Nested Network Moblty Mohammed Babker Al Mohammed, Asha Hassan.

More information

Optimal Workload-based Weighted Wavelet Synopses

Optimal Workload-based Weighted Wavelet Synopses Optmal Workload-based Weghted Wavelet Synopses Yoss Matas School of Computer Scence Tel Avv Unversty Tel Avv 69978, Israel matas@tau.ac.l Danel Urel School of Computer Scence Tel Avv Unversty Tel Avv 69978,

More information

Petri Net Based Software Dependability Engineering

Petri Net Based Software Dependability Engineering Proc. RELECTRONIC 95, Budapest, pp. 181-186; October 1995 Petr Net Based Software Dependablty Engneerng Monka Hener Brandenburg Unversty of Technology Cottbus Computer Scence Insttute Postbox 101344 D-03013

More information

Application of VCG in Replica Placement Strategy of Cloud Storage

Application of VCG in Replica Placement Strategy of Cloud Storage Internatonal Journal of Grd and Dstrbuted Computng, pp.27-40 http://dx.do.org/10.14257/jgdc.2016.9.4.03 Applcaton of VCG n Replca Placement Strategy of Cloud Storage Wang Hongxa Computer Department, Bejng

More information