Fair Intelligent Congestion Control Resource Discovery Protocol on TCP Based Network 1

Size: px
Start display at page:

Download "Fair Intelligent Congestion Control Resource Discovery Protocol on TCP Based Network 1"

Transcription

1 Far Intellgent Congeston Control esource Dscovery Protocol on TCP Based Network 1 Doan B. Hoang 2, Qng Yu, Mng L, and Davd Dagan Feng Department of Computer Systems, Faculty of Informaton Technology, Unversty of Technology, Sydney, NSW 2007, Australa e-mal: [dhoang, mngl]@t.uts.edu.au School of Informaton Technologes, Unversty of Sydney, NSW 2006, Australa e-mal: [qng, feng]@t.usyd.edu.au Abstract Today s Internet only provdes best-effort servce for all traffcs. The network s not able to guarantee the qualty of servce requred by an applcaton that demands more strngent response n terms of delay, jtters, bandwdth and etc. It s well accepted that, the deployment of QoS-aware technologes s a key factor for the contnued success of the Internet. In ths paper, we propose a Far Intellgent Congeston Control esource Dscovery (FICCD) protocol on TCP based network whereby a mechansm s employed at core routers to determne avalable network resources and convey ths nformaton to edge routers. At the edge routers, an ntellgent control algorthm s employed to assst the TCP to maxmze ts traffc over the underlyng network. The key deas are to ntegrate avalable network resources n estmatng connectons far share; to create feedback control loops between edge routers; to employ a specal esource Dscovery (D) packet to collect and convey en route router state nformaton; and to employ ntellgent algorthms to match a TCP connecton s sendng rate to the rate at whch the underlyng network can support. We demonstrate that FICCD protocol s effectve, far, flexble and can be easly extended for QoS control of the future Internet. Keywords TCP, Congeston Control, Explct Wndow Adaptaton, Acknowledgment Bucket Control, Feedback Control Loop 1. INTODUCTION Today s Internet only provdes best-effort servce for all traffc. Traffc s processed as quckly as possble but there s no guarantee as to tmelness or actual delvery. The network makes no attempt to dfferentate ts servce response between the traffc streams generated by concurrent users of the network. Ths means that the network s not able to guarantee the level of servce requred by an applcaton that demands more strngent response n terms of delay, jtters, bandwdth, etc. A smple applcaton such as vdeoconference over the Internet wll quckly demonstrate the problems wth real-tme applcatons. A person takng part n a vdeoconference wll mmedately realze that the qualty of the mages s not as good as expected, the flow of mages s not qute smooth and the synchronzaton between voce and mages s far from perfect. The evoluton of the Internet s at a turnng pont. It s well accepted that, n ts evoluton n the 21st century, the deployment of Qualty of Servce (QoS) aware technologes s a key factor for the contnued success of the Internet. Some advances n ths drecton have already been provded wth the ntroducton of the Internet augmented archtectures wth QoS supports: the Internet Protocol Verson 6 (IPV6) [2], the Integrated Servces [3-4] and Dfferentated Servces [5-6] by the IETF. Furthermore, the eservaton Protocl (SVP) [7] and the Mult Protocol Label Swtchng (MPLS) technque [8] provde addtonal support n terms of sgnalng and traffc engneerng. At the nfrastructure level, the deployment of Wavelength Dvson Multplexng (WDM) systems [9-10] creates a huge transport capacty, whch allevates the bandwdth demand of multmeda applcatons over the Internet. Despte these advances, many ssues stll reman to be solved. Among them, ssues concernng admsson control and resource allocaton polces n the network, adequate user sgnalng protocols, network management systems for large numbers of users and securty requre closer attenton. It s expected that, havng n place the adequate nfrastructure and protocols, hgh qualty real-tme applcatons can run on the Internet wthout sgnfcant nterference from conventonal data traffc. 1 Ths work was supported n part by AC Grant. 2 The correspondng author

2 Our approach s motvated by recent technologcal changes and the need for the adequate nfrastructure and protocols towards the QoS-aware network. We beleve that for effectve QoS-aware technologes, QoS ssues have to be tackled at several levels as shown n Fgure 1.1. At the network level, relevant nformaton concernng the network operatonal condtons and avalablty of resources of the underlyng network s essental to avod network congeston. At the transport level, relevant nformaton concernng the end systems s crucal for far bandwdth sharng and end-to-end QoS per connecton. At the applcaton level, relevant nformaton concernng the end applcatons are necessary to provde approprate QoS for the applcaton. We also beleve that feedback mechansm s essental for QoS control n networkng. By couplng wth relevant protocol for transportng necessary feedback nformaton as well as control actons amongst partcpatng enttes, we can make congeston control decsons, resources allocaton decsons, QoS decsons more ntellgent. Fgure 1 Feedback Mechansm for QoS Control In ths paper we develop a feedback control loop at the network layer to provde the transport protocol wth necessary nformaton for optmsng ts operaton. Several algorthms are employed to maxmse the TCP sendng rate of a connecton over the avalable bandwdth of the underlyng network. In partcular, we propose Far Intellgent Congeston Control esource Dscovery protocol (FICCD) on TCP based network n whch a mechansm s employed at core routers to determne avalable network resources and convey ths nformaton to edge routers. At the edge routers, an ntellgent control algorthm s employed to assst the TCP to maxmze ts traffc over the underlyng network. The key deas are to ntegrate avalable network resources n estmatng connectons far share of network resource; to create feedback control loops between edge routers; to employ a specal esource Dscovery (D) packet to collect and convey en route router state nformaton; and to employ ntellgent algorthms to match a TCP connecton s sendng rate to the rate at whch the underlyng network can support. We demonstrate that the framework can sgnfcantly mprove n throughput, farness, and packet loss rate for end-to-end TCP connectons. More mportantly, our protocol s transparent to TCP, requres no modfcatons to current TCP mplementatons and can be easly extended for QoS control of the future Internet. The paper s organzed as follows. We begn n secton 2 by brefly descrbng some related work. Our proposal s presented n secton 3. Secton 4 descrbes smulaton scenaro, smulaton envronment and smulaton parameters. The smulaton results and evaluaton are presented n secton 5. Some concludng remarks and drecton of future work are gven n secton ELATED WOK A computer network typcally uses store-and-forward routng to transfer data packets between users at geographcally dstrbuted nodes. Packets generated by a source node are delvered to ther destnaton by routng them va a sequence of ntermedate nodes. The traffc flowng through an ntermedate node depends upon the number of source-destnaton pars that are routed through that node and the rates at whch these sources ntroduce packets nto the network. If the source rates are ncreased wthout constrant, queues of packets watng to be

3 routed buld up at bottleneck nodes. Eventually, the bufferng capacty of these nodes s exceeded and packets are dropped, resultng n low throughput and hgh delay. Congeston control mechansms attempt to avod such breakdown by mposng constrants on the sources. Two types of constrant are often used. In rate-based congeston control, a lmt s placed on the rate at whch a source can send packets. In wndow-based congeston control, at any nstant there s a lmt to the number of outstandng packets at the source, but there s no constrant on the rate at whch packets can be sent. In ATM network [11], the Avalable Bt ate (AB) servce s adopted for transportng best-effort data traffc wth no delay guarantees; AB rate-based congeston control attempts to mnmze the cell loss rato, and provde mnmum cell rate guarantees through the closed loop feedback control mechansm. The network provdes feedback to the sources when network load changes, and the sources adjust ther transmsson rates accordngly. AB congeston control handles congeston effectvely. Many studes have demonstrated that ATM AB servce can provde low-delay, farness, and hgh throughput, and can handle congeston effectvely nsde the ATM network. However, network congeston s not really elmnated but rather t s pushed out to the edge of the ATM network. Packets from TCP sources competng for the avalable ATM bandwdth are buffered n the routers or swtches at the network edges, causng severe congeston, degraded throughput, and unfarness. In IP networks, Floyd [12] proposed to modfy TCP slghtly to nclude an Explct Congeston Notfcaton (ECN) from routers to the sources to trgger a wndow sze reducton dentcal to that caused by the fast retransmt fast recovery mechansm of TCP-eno, yet wthout droppng packets. By combnng explct notfcaton wth ED, the performance of both delay-senstve (telnet-lke) and delay-nsenstve (ftp-lke) traffc can be mproved. However, because of ts bnary feedback, ECN cannot avod wndow and network oscllatons. These can negate the gan n network performance. Other approaches such as Tr-S [13] and TCP-Vegas [14] attempt to estmate the bandwdth-delay product for each TCP connecton and adjust the wndow sze based on ths estmate. However, these schemes ntroduce complexty n the end-system and requre extensve modfcatons to current TCP mplementatons. Gerla et al. proposed a feedback-based algorthm (BA-TCP) [15-16] at network layer to convey the roundtrp propagaton delay and avalable bandwdth nformaton for each TCP connecton. The end hosts use ths nformaton to adjust ther congeston wndow. However, knowledge of the TT s usually not avalable at current router. The scheme also needs to modfy current TCP mplementaton by addng one state varable to store the round-trp propagaton delay and advertsng the mnmum of the recever s buffer space and the avalable bandwdth-delay product. Hjalmtysson [17] descrbed a control-on-demand model for programmable networks. Ths model allows the nstalled control programs to explot lower-level facltes, n partcular hardware facltes. Through flterng and frame peekng, control programs can nspect flows or aggregate or control stream of Internet and then extract the requred nformaton for ntellgent control. Wth ths approach, one does not have to modfy exstng router n anyway snce the control servce can be downloaded or njected to the router on the fly wthout nterruptng ts operaton. Harsson and Kalyanaraman [18] proposed edge-to-edge feedback-based control algorthm at network layer to regulate the aggregate traffc between each edge-par. Each control loop creates a vrtual lnk and the exchange of control packets operates on a per-edge-to-edge vrtual lnk bass. A new set of congeston control technques s requred to construct vrtual lnks, whch break up congeston at nteror nodes and dstrbute the smaller congeston problems across the edge nodes. However, the scheme s requrement on routers s hgh. It may be unrealstc to expect routers to devote much of ther resources to exercsng the algorthm and handng the admsson control. 3. FAI INTELLIGENT CONGESTION CONTOL ESOUCE DISCOVEY POTOCOL In ths secton, we provde detals of the proposed framework and show how the network operates under the proposed framework. We consder smplfed node, local access network and wde area network archtectures for current Internet Infrastructure (fgure 2).

4 The am of FICCD protocol s twofold. Frstly, t ams to develop a feedback loop n WAN regon between the edge routers to convey nformaton concernng avalable network resources and network condtons from wthn the network. We have deployed a smlar scheme, the Far Intellgent Congeston Control (FICC) [19-20], successfully for ATM s AB congeston control. In ths paper the nnovatve aspect s n nvestgatng how such a feedback scheme can be useful over the Internet. Secondly, t ams to maxmze a TCP connecton throughput by matchng ts sendng rate to the rate at whch the underlyng network can support. We am to establsh a feasble framework over whch explct feedback nformaton can be conveyed. Qualty of Servce (QoS) and congeston control mechansms can then be deployed at the edge devces based on the feedback nformaton. Our ntal mplementaton s concerned wth ndvdual connectons to see how TCP operates under ths scheme. Our next step s to nvestgate the scalablty of the scheme by applyng t on a per-class bass as specfed by the dfferentated servces code pont (DSCP) n DffServ. Bascally the control loop operates as follows. A specal resource dscovery (D) packet s generated at source edge router for each flow proportonally based on the packet arrval rate. The D-packet wll carry the arrval packet rate (or ts estmate) n ts AP feld. Each core router along the path drectly updates feedback nformaton concernng the network condtons n the D packets when they pass n the forward or backward drecton. The destnaton edge router sends back the D-packets. When the source edge router receves Backward D-packet, t passes the feedback nformaton to an algorthm that s responsble for matchng the TCP sendng rate to the expected rate supported by the network. Hosts Edge outer Border outers Edge outer Hosts L- E- B- C- B- E- L- Leaf outer LAN WAN LAN Leaf outer D(Q S 1,,..) D(Q 1 1, ) D(Q k 1, ) 3.1 Edge outer behavors Fgure 2 Internet General Model and FICCD Control Loop As far as the FICCD s concerned, an edge router s responsble for estmatng packet arrval rate for ts connecton, ntatng and mantanng the D control loop, collectng the feedback nformaton, and exercsng control algorthm for coordnatng control between TCP and the underlyng network Forwardng behavors In the forward drecton, the source edge router classfes the ncomng packets nto flows and the arrval packet rate of each flow denoted as AP (t). Snce an exact computaton of the arrved rate of each flow s hardly feasble, a per-flow rate estmate s updated upon the recepton of every packet usng exponental AP (t )

5 averagng formula as n CSFQ [21]. Usng an exponental weght gves more relable estmaton for bursty traffc, even when the packet nter-arrval tme has sgnfcant varance. If we ndcate the arrval tme of the k-th packet of flow as T K and ts length as, the new estmate of AP (t ) can be computed as follows: AP new ( t ) l k ( t ) k k k T / K l T / K old = (1 e ) k + e AP (1) T K Where represents the k-th sample of the nterarrval tme of flow,.e., T K k ( k 1) T = and K s a t t constant. D packets are generated at source edge router for each flow proportonally based on the packet arrval rate n Eqn. (1). D-packet wll carry the arrval packet rate n ts AP feld AP (t ) AP (t ) Edge outer Packet Buffer Packets ACKs Control ACK Buffer Explct Wndow Adaptaton Normal ACK Bucket Control Backwardng behavors Fgure 3 Edge outer model In the backward drecton, two approaches (Explct Wndow Adaptaton, ACK Bucket Control) can be employed at source edge router to convey such feedback nformaton concernng the network condtons to TCP source. The Explct Wndow Adaptaton approach controls the maxmum recever wndow (WND) at edge router, whch acts as an upper bound on the TCP congeston wndow (CWND) varable at the TCP source, thus effectvely controls the TCP sender rate by matchng t wth the feedback nformaton concernng the network condtons at the network layer. The ACK Bucket Control approach ams to match the TCP sender rate wth the avalable feedback resources by wthholdng the Acknowledgments at the edge router. We have recently proposed a Far Intellgent Explct Wndow Adaptaton (FIEWA) scheme and a Far Intellgent ACK Bucket Control (FIABC) scheme for each of the two approaches [22-23]. The essental deas of these schemes are summarzed below Far Intellgent Explct Wndow Adaptaton The objectve of Far Intellgent Explct Wndow Adaptaton (FIEWA) s to match the sum of the wndows of all actve TCP connectons sharng the buffer at the edge router to the effectve network delay-bandwdth product, thus avodng packet losses whenever possble. Two key components are ncluded n FIEWA: () A mechansm to sgnal wndow updates from the network to the source, and () a scheme at the bottleneck pont to estmate the avalable bandwdth based on the congeston state of the network. The former can be accomplshed by allowng the network edge elements to modfy the recever s advertsed wndow feld carred by TCP acknowledgements from the destnaton to the source. The latter problem, however, s sgnfcantly more dffcult. Frst, when there are multple bottlenecks on the path of a TCP connecton, the wndow estmaton algorthms at these bottlenecks may nteract n undesrable ways. Second, estmatng the avalable bandwdth of an output lnk of a bottleneck element s rather dffcult. Fnally the advantages of the scheme must be compared aganst those of mplct (packet dscard) schemes of comparable complexty.

6 For the specfc network envronment under consderaton, the stuaton s much smpler. Frst, we assume that the congeston between source edge router to destnaton edge router s well control; the only bottleneck n the path of the TCP connectons occurs at the source edger router. Second, both the bandwdth avalable n the edge router, and the delay through t, reman relatvely steady over short tmescales, ndependent of the number of TCP connectons transported over t. Ths makes t easer for the edge router to estmate the avalable bandwdthdelay product for each TCP connecton. Bascally, the FIEWA works as follows. FIEWA sends explct feedback nformaton n the form the recever s advertsed wndow feld of the returnng TCP acknowledgments to TCP sources to adjust ther wndow szes. If the current value n the recever s advertsed wndow, whch s set by the destnaton system, exceeds the feedback value computed n the edge router, the recever s advertsed wndow s marked down to the feedback value. The computed feedback value s a functon of the free buffer space at the edge router. Let Q denote the empty buffer space at tme t when a returnng ACK arrves at an edge e ( t) = BufferSze Q( t) devce, where BufferSze s the total buffer space and Q (t) the total buffer occupancy at tme t. Let W denote the r (t) value n the recever s advertsed wndow feld seen n the ACK. The FIEWA algorthm computes a target wndow sze for the TCP connecton as a functon of the avalable buffer, that s f ( Q. Ths computed value e ( t )) s then used to mark down the recever s advertsed wndow feld n the acknowledgement. Snce settng the wndow sze smaller than the maxmum segment sze (MSS) negotated durng connecton establshment can lead ' to starvaton and deadlocks, a mnmum wndow sze of MSS s enforced. Thus, the feedback value, W, used r ( t ) to set the recever s advertsed wndow feld, s computed at the edge devce as ' W r ( t ) = MAX ( MIN ( W r ( t ), f ( Q e ( t ))), MSS ) (1) The dffcult task n such an algorthm s to desgn the feedback functon f ( Q. The dynamcs of the e ( t )) system depend heavly on ths feedback functon. The goal of the functon s to provde all TCP connectons wth smlar feedback, and as a result they all operate wth equal wndows. Snce a lnear functon s smple and stll effectve, t s employed n our feedback functon. Explctly, the wndow for each connecton s set to the avalable buffer space multpled by a fracton whch determnes the buffer occupancy n steady state. That s f ( Q ( t )) = DPF ( t ) * Q ( t ) = DPF ( t ) * ( BufferSze Q ( t )) (2) Far Intellgent ACK Bucket Control e e The objectve of Far Intellgent ACK Bucket Control scheme (FIABC) s to match the TCP sender rate wth the avalable bandwdth n underlyng network by wthholdng the acknowledgment packets at edge router, so that TCP source cannot send packets more than that can be handled by the underlyng network. We choose to wthhold ACK packets at edge router based on the followng consderatons. Snce the essental nformaton contaned n an acknowledgment s only a sequence number, ths acknowledgment buffer s really only a lst of numbers. To dfferentate ths type of bufferng from the tradtonal packet buffer, we wll call t acknowledgment bucket. By controllng the flow of acknowledgments from the destnaton to the source, we can practcally elmnate the large packet buffer and stll acheve the same output rate as f we have unlmted packet buffer at the network edges. The acknowledgment bucket s analogous to the storage of permts n the well-known leaky bucket scheme. TCP acknowledgments serve as permts that allow the edge devce to request packets from the source. When the acknowledgment bucket s empty, the edge cannot request any more packets. The acknowledgment bucket gets flled up accordng to TCP dynamcs and s draned out accordng to the avalable bandwdth n the underlyng network. The acknowledgment bucket serves as a translator, whch transforms the avalable bandwdth command nto a sequence of TCP acknowledgments whose effect s to have the TCP source sent data no faster than the avalable bandwdth. The dffcult task n such an algorthm s to determne the forwardng rate for the acknowledgments. Snce the edge devce serves as a vrtual source for the underlyng network, our approach determnes the acknowledgment packet releasng by comparng the request from underlyng network wth current edge devce buffer length. In detal, our approach s to estmate the amount of data requested by underlyng network from last ACK release at edge devce correspondng the avalable bandwdth denoted as AB. That s, the amount of data

7 for each connecton requred by underlyng network at tme t, denoted as AB t )*( t t n steady state. ( lastackrelease Underlyng ( t ), s set to ) multpled by a queue control functon DPF that determnes edge router buffer occupancy Underlyng ( t lastackrel ease ) = DPF ( t ) * AB( t ) * ( t t ) (1) Note that the amount of data requested by the underlyng network Underlyng ( t ) from last ACK release must be smaller than the advertsement wndow sze n the latest ACK packet receved by edge router. t ) = Mn ( Underlyng t ), WND) (2) Underlyng ( We denote Q t ) the packet queue length at tme t at edge devce. Snce the packet queue at edge devce ( servces as a vrtual source for the underlyng network, FIABC schedule the ACK releasng from the ACK Bucket by comparng currently what the underlyng network request the packets from the edge devce Underlyng t ) wth currently what edge devce can provde n ts packet queue Q t ). Whenever Underlyng t ) > Q t ), the edge devce wll release an acknowledgment from the ACK Bucket. 3.2 Core outer behavors We employed the Far Intellgent Congeston Control (FICC) scheme [18] at each core router to estmate a far share of bandwdth for competng TCP connectons, calculate avalable network capacty and feedback relevant nformaton to the edge router. It has been demonstrated that FICC s smple, robust, and effectve congeston control algorthm. Importantly, FICC s also able to allocate bandwdth farly among ts connectons. Essentally, n order to estmate the current traffc generaton rate of the network and allocate t among connectons farly, a Mean Allowed Packet ate (MAP) s kept at each core router. MAP = MAP + β * (AP MAP) Where AP s the value of the current arrval packet rate carred n the AP feld of the arrvng forward D-packet. MAP represents an estmate of the average load passng through the router at the current tme. When the network operates at the acceptable level, the correspondent MAC s regarded as the optmal packet rate for each flow. In Far Intellgent Congeston Control mechansm, the network s expected to work at the target operatng pont. The target operatng pont adopted n ths scheme s a pre-set Buffer Utlzaton ato (BU), whch means that the optmal control s to keep the buffer utlzaton at an optmal level. The motvaton behnd ths dea s to make effcent use of the buffer capacty. D (AP, E, CI) D (AP, E, CI) D (AP, E, CI) D (AP, E, CI) ( ( ( ( ( D Data Control Element (CE) FE Q6/w 6 : Premum Q5/w 5 : Platnum(AF4) Q4/w 4 : Gold (AF3) Q1/w 1 : Best Effort(DE) C n Fgure 4 Core outer model

8 To calculate the expected rate (E) based on the queue length at core router, a lnear queue control functon DPF s employed n our scheme. The basc characterstcs of the functon are that t has a value equal to 1 when the queue length s target queue length Q0, and a value less/larger than 1 when the queue length s larger/less than Q0. The larger/smaller the queue length, the smaller/larger the factor to push forward the network to the target operatng pont. Snce queue s bult up and draned out contnuously, queue control functon s desred to perform contnuous control to produce proper effect on the queue fluctuaton and smooth the computed E values. The pseudocode of FICCD s shown n table 1. efer to [1, 19] for further descrpton of the algorthm. Forward Table 1. Descrpton of the Core outer s Algorthm D (AP, E, CI) If Queue > Q0 True If AP < MAP True MAP=MAP+b*(AC-MAP) D(AC,E,CI) C O E O U T E D(AC,E,CI) False MAP=MAP+b*(AP-MAC) If E > DPF*MAP DPF=(a-1)*(Q0-QueueLength)/Q0+1 False If Queue > Q0 D(AP,E,CI) C O E O U T E True True E=DPF*MAP DPF=(Buffersze- ueuelength)/(buffersze-q0) 4 SIMULATION SETUP Backward We use ns (verson 2) [24] network smulator to evaluate the proposed FICCD protocol. Ns s a dscrete event smulator for network research. It provdes substantal support for TCP, router queung mechansms, and varous topologes. New components edge router, core router and new protocol FICCD were added and compled nto ns. The smulaton performs wth TCP applcatons runnng over an IP network. Peer-to-Peer confguraton [Fgure 4] s employed n the smulaton. There are 10 sources sendng data to 10 dstnct destnatons through a sngle bottleneck lnk between 11 and 12 wth buffersze of 200pkts, propagaton delay of 20 ms and bandwdth of 10 Mbts/s. Source 0 sends data to the destnaton 14 (on path 10, 11, 12, 13) has propagaton delay of 10 ms and bandwdth of 10 Mbts/s. All the other lnks are wred wth propagaton delay of 2 ms and bandwdth of 10Mbts/s. The pont we want to look at s the far share of bandwdth among multple connectons wth dfferent TT sharng a bottleneck lnk. Each source has an nfnte data to send. The sze of data packets s 4Kbyte. The smulaton s run for 200 seconds. TCP clock granularty s set to 0.3 seconds and the recever s wndow sze s set to128 Kbytes. The TCP verson used n our smulatons was TCP eno, whch ncludes fast retransmsson and fast recovery.

9 10Mbt/s, 10ms 100Mbt/s, 10ms 1.5Mbt/s, 20ms 100Mbt/s, 10ms 10Mbts,2ms Fgure 4 Peer to peer confguraton 5. SIMULATION ESULTS AND ANALYSIS 10Mbts,2ms In ths secton we present the smulaton results for FICCD (denoted as FICCD (FIEWA) and FICCD (FIABC) ), smple Droptal (denoted as Droptal ), ED (denoted as ED ) and ECN wth ED (denoted as ED+ECN ) for the purpose of comparson and dscusson. We evaluate the smulaton results n terms of the goodput, the sender sequence number at TCP source, the packet queue length, and the droppng total at the bottleneck core router. 5.1 Packet Queue Length The man reason for TCP performance degradaton s due to the overflow of buffers at bottleneck router. An mportant attrbute of the TCP congeston control mechansm s that t does not assume any explct sgnalng of congeston state from the underlyng network for. It nfers the congeston state of the network mplctly: the arrval of acknowledgements (ACKs), tmeouts, and recept of the duplcate ACKs. As a result, the wndow-based congeston control mechansms of TCP may nteract wth the underlyng network n undesrable ways, causng severe congeston, degraded throughput, and unfarness. Queue Length QueueLength Dropt al FICCD+FIABC ED+ECN FICCD+FIEWA ED Max. Avg. Mn Tme(M llseconds) Dr opta l FICCD(FIEWA) FICCD(FIABC) ED+ECN ED Fgure 5 Bottleneck Queue Length As shown n Fgure 5, wth Droptal, the packet queue length grows beyond the maxmum packet buffersze of the bottleneck router, and packets have to be dropped, packet retransmsson occurs when the TCP source becomes aware of the loss of packet. Wth ED, two fxed thresholds (hgh threshold and low threshold) are used to detect congeston, dfferent droppng polces are employed for the congeston perod when

10 the queue length s greater than the hgh threshold and the congeston perod when the queue length s between the two thresholds. Both Droptal and ED use droppng mechansm at router to ndcate the congeston and control TCP sendng rate mplctly. The queue length varaton s between bytes to bytes for Droptal and between bytes t bytes for ED. By combnng ECN wth ED, the bottleneck router can set the congeston bt n the ACK return back to TCP, thus provdng TCP an explctly nformaton to controllng ts sendng rate. ather than usng a fxed queue threshold to arbtrarly dvde a network nto congeston and noncongeston states, our FICC algorthm ams for a target operatng pont where the router queue length s at an optmal level for good throughput and low delay, and where the allocaton s optmal for each connecton. The queue control functon encourages traffc sources f the target operatng pont s not reached, and dscourages the sources f the swtch operates beyond ts target operatng pont. As shown n Fgure 5, FICCD mantans the queue level at the target level around bytes wth narrow range ([ bytes, bytes]). 5.2 Goodput We use the current acknowledged bytes dvded by the smulaton duraton tme to calculate the goodput. Fgure 6 shows the goodput for each flow durng ntervals seconds. The goodput shown n ths fgure has been normalzed by the bottleneck far share (.e., by 0.15 Mbts/s). Therefore, a goodput of 0.15 represents a data throughput of 0.15 Mbts/s. The fgure shows an unfar allocaton of bandwdth when other schemes are used. In contrast, the throughput obtaned by FICCD s far n that all TCP connectons are allocated roughly the same amount of bandwdth. We notce the average effectve throughput shows a slghtly lower than the estmated value of 0.15Mbts/s. Ths manly comes from the neffcency of the TCP slow-start, the overhead costs concernng packet headers, the paddngs, and the bandwdth requred for D packets. The farness property of FICCD comes from three man factors: Frstly, at each core router, FICCD estmates accurately the far share of ts connectons. Secondly, FICCD bulds n an oversell feature that allows unconstraned connectons to take up the leftover bandwdth that cannot be taken up by constraned connectons. Ths allows the bandwdth to be shared among unconstraned connectons farly. Thrdly, FICCD does not allow the operatng pont to dverge far wary from ts stable pont. Ths results n far share and small devatons n the router queue length, n packet delay and packet delay varaton. Goodput Sender sequence number Total Sequence number Dropt al FICCD(FIABC) ED+ECN FICCD(FIEWA) ED Tme(M Illseconds) Dr opta l FICCD(FIEWA) FICCD(FIABC) ED ED+ECN Fgure 6. Goodput Fgure 7. Sender Sequence Number 5.3 TCP Sender Sequence Number We use TCP sender sequence number versus tme to pant a pcture of TCP sendng rate as shown n Fgure 7. Comparatvely, t s clear that the TCP sendng rate n FICCD mechansms s less than other cases before the packet droppng, whch means that the burstness of traffc s reduced based on explct feedback rate n D packet n FICCD. The feedback loop n our schemes s explct and effectve n that the TCP source rate s drectly controlled to adapt to the avalable bandwdth, rather than progressvely adjusted through TCP wndow flow control, whch reles on packet drop as ndcaton of network congeston.

11 Dropng Total ED+ECN ED FICCD(FIABC) FICCD(FIEWA) Dr op Ta l Fgure 8 Droppng Total 5.4 Dropng Total We also show the droppng total for each flow n Fgure 8. It shows clearly that wth our scheme there are no packets droppng at bottleneck router. Wth Droptal, the packet queue length frequently grows above the buffer lmt and hence packets are dscarded; the droppng total (50000 bytes) s largest compared wth other schemes. ED and ED+ECN use dfferent droppng polces dependng on the congeston states. Overall, ED+ECN performs slghtly better than ED wth a droppng total of bytes, comparng wth bytes n ED. 6. CONCLUSION AND FUTUE WOK Ths paper proposes the Far Intellgent Congeston Control esource Dscovery (FICCD) protocol for the purpose of mprovng end-to-end TCP performance by controllng the congeston and allocatng far share of bandwdth to competng TCP traffcs. The essental dea of the Far Intellgent Congeston Control esource Dscovery protocol s the ntroducton of a feedback loop and ts assocated protocol to collect relevant nformaton about the underlyng network between a source and a destnaton edge routers par; and the use of the nformaton to regulate the TCP congeston control and maxmze ts performance. The sgnfcant contrbutons of FICCD protocol nclude: Integratng of network resources, such as avalable bandwdth, avalable buffer, queung delay and jtter to estmate far share of network resources among competng traffc, Allowng possble ntegraton of LAN and WAN together to provde end-to-end QoS for applcaton by allowng the mappng of LAN request to approprate QoS of WAN supported by feedback control mechansm, Permttng an admsson control mechansm to make use of the feedback nformaton to provde adequate level of control. The performance evaluaton s based on the goodput, farness, buffer requrement, etc. The smulaton results show that our scheme s effectve n mprovng goodput, achevng farness, and mnmzng packet loss rate for end-to-end TCP connectons. Importantly, our framework s transparent to TCP, requres no modfcatons to the current TCP mplementaton and can be easly extended for QoS control of the future Internet. Currently, we are evaluatng the overheads ncurred by the feedback loop and D packets. We are also nvestgatng the use of FICCD for congeston and QoS control n the Dfferentated Servces archtecture. The goal s to apply FICCD not on a per flow bass but on a per DSCP or class bass so as to preserve the scalablty of Dffserv.

12 EFEENCES 1) Q. Yu, Mng L, Hoang, D. B. and Davd Dagan Feng, "Far Intellgent Feedback Mechansm on TCP Based Network", to appear n the proceedngs of the 2002 Internatonal Conference on Internet Computng (IC 2002), Las Vegas, Calforna, USA, June ) Internet Protocol Verson 6 (IPV6). 3). Braden, D. Clark, and S. Shenker, Integrated Servces n the Internet Archtecture: an overvew, IETF FC 1633, June ) D. Clark, S. Shenker, and L. Zhang, Supportng real-tme applcatons n an ntegrated servces packet networks: Archtecture and mechansm, ACM SIGCOMM, Baltmore, ) I., Stoca, L. Zhang, A model for servce dfferentaton n the Internet, NOSSDAV 98, ) D. Clark, W. Fang, Explct allocaton of best effort packet delvery servce, IEEE/ACM Transactons on Networkng, Vol. 6, Issue 4, ) L. Zhang, et al., SVP: A new resource reservaton protocol, IEEE Network, Vol. 7, Issue 5, pp. 8-18, ) E. osen, A. Vswanathan, and. Callon, Multprotocol Label Swtchng Archtecture, IETF FC 3031, January ) A.S. Acampora, M.J. Karol, An overvew of lghtwave packet networks, IEEE Network pp , Jan, ) E. Modano, WDM-based packet networks, Magazne of IEEE Communcaton, pp , Mar, ) W. Stallngs, Hgh-Speed Networks: TCP/IP and ATM Desgn Prncples, Prentce-Hall, ) S. Floyd, TCP and explct congeston notfcaton, Computer Communcaton evew, vol. 24, no. 5, pp. 8-23, October ) Z. Wang and J. Crowcroft, A new congeston control scheme: slow start and search (Tr-S), Computer Communcaton evew, vol. 21, no. 1, pp , January ) L. S. Brakmo, L. L. Peterson, TCP Vegas: End to end congeston avodance on a global Internet, IEEE Journal on Selected Areas n Communcatons, vol. 13, no. 8, pp ) M. Gerla, W. Weng,. Lo Cgno, BA-TCP: A bandwdth aware TCP for satellte networks, Proceedngs of IEEE ICCCN 99, Boston, Massachusetts, October ) M. Gerla, W. Weng;.Lo Cgno,, Bandwdth feedback control of TCP and real tme sources n the Internet, Global Telecommuncatons Conference, 2000 (GLOBECOM '00). IEEE, Volume: 1, pp: , ) G. Hjalmtysson, Control on-demand: An Effcent Approach to outer Programmablty, IEEE Journal on Selected Areas n Communcatons, Vol. 17, No. 9, pp , Sept ) Davd Harrson, Shvkumar Kalyanaraman, "Edge-to-Edge Traffc Control for the Internet: Concepts and Archtecture," PI ECSE Networks Laboratory Techncal eport, ECSE-NET January ) D. B. Hoang, and Z. Wang, A Far Intellgent Congeston Control for ATM, Techncal eport 224, ISBN , School of Informaton Technologes, Sydney Unversty, Australa, December ) Hoang, D. B., and Yu, Q., "Performance of the Far Intellgent Congeston Control for TCP Applcatons over ATM Networks", Proceedngs of the Second Internatonal Conference on ATM (ICATM'99), Colmar, France, June 1999, pp ) I. Stoca, S. Shenker, and H. Zhang, Core-Stateless Far Queueng: Achevng Approxmately Far Bandwdth Allocatons n Hgh Speed Networks, In Proceedngs of the ACM SIGCOMM 98, Vancouver, Canada, September ) Yu, Q., and Hoang, D. B., "A Coherent Approach to Cooperaton between TCP and ATM Congeston Control," submtted to specal ssue of the Transactons of the Socety for Computer Smulaton -Part B on Performance Modelng and Smulaton of ATM Systems and Networks. Y 23) Yu, Q., and Hoang, D.B., "Far Intellgent Explct Wndow Adaptaton," IC'2001, Las Vegas, Calforna, USA, pp ) ns-2, Network Smulator(verson.2). LBL, UL:

PAS: A Packet Accounting System to Limit the Effects of DoS & DDoS. Debish Fesehaye & Klara Naherstedt University of Illinois-Urbana Champaign

PAS: A Packet Accounting System to Limit the Effects of DoS & DDoS. Debish Fesehaye & Klara Naherstedt University of Illinois-Urbana Champaign PAS: A Packet Accountng System to Lmt the Effects of DoS & DDoS Debsh Fesehaye & Klara Naherstedt Unversty of Illnos-Urbana Champagn DoS and DDoS DDoS attacks are ncreasng threats to our dgtal world. Exstng

More information

VoIP Playout Buffer Adjustment using Adaptive Estimation of Network Delays

VoIP Playout Buffer Adjustment using Adaptive Estimation of Network Delays VoIP Playout Buffer Adjustment usng Adaptve Estmaton of Network Delays Mroslaw Narbutt and Lam Murphy* Department of Computer Scence Unversty College Dubln, Belfeld, Dubln, IRELAND Abstract The poor qualty

More information

benefit is 2, paid if the policyholder dies within the year, and probability of death within the year is ).

benefit is 2, paid if the policyholder dies within the year, and probability of death within the year is ). REVIEW OF RISK MANAGEMENT CONCEPTS LOSS DISTRIBUTIONS AND INSURANCE Loss and nsurance: When someone s subject to the rsk of ncurrng a fnancal loss, the loss s generally modeled usng a random varable or

More information

Network Services Definition and Deployment in a Differentiated Services Architecture

Network Services Definition and Deployment in a Differentiated Services Architecture etwork Servces Defnton and Deployment n a Dfferentated Servces Archtecture E. kolouzou, S. Manats, P. Sampatakos,. Tsetsekas, I. S. Veners atonal Techncal Unversty of Athens, Department of Electrcal and

More information

Traffic State Estimation in the Traffic Management Center of Berlin

Traffic State Estimation in the Traffic Management Center of Berlin Traffc State Estmaton n the Traffc Management Center of Berln Authors: Peter Vortsch, PTV AG, Stumpfstrasse, D-763 Karlsruhe, Germany phone ++49/72/965/35, emal [email protected] Peter Möhl, PTV AG,

More information

The Development of Web Log Mining Based on Improve-K-Means Clustering Analysis

The Development of Web Log Mining Based on Improve-K-Means Clustering Analysis The Development of Web Log Mnng Based on Improve-K-Means Clusterng Analyss TngZhong Wang * College of Informaton Technology, Luoyang Normal Unversty, Luoyang, 471022, Chna [email protected] Abstract.

More information

Efficient Bandwidth Management in Broadband Wireless Access Systems Using CAC-based Dynamic Pricing

Efficient Bandwidth Management in Broadband Wireless Access Systems Using CAC-based Dynamic Pricing Effcent Bandwdth Management n Broadband Wreless Access Systems Usng CAC-based Dynamc Prcng Bader Al-Manthar, Ndal Nasser 2, Najah Abu Al 3, Hossam Hassanen Telecommuncatons Research Laboratory School of

More information

VRT012 User s guide V0.1. Address: Žirmūnų g. 27, Vilnius LT-09105, Phone: (370-5) 2127472, Fax: (370-5) 276 1380, Email: info@teltonika.

VRT012 User s guide V0.1. Address: Žirmūnų g. 27, Vilnius LT-09105, Phone: (370-5) 2127472, Fax: (370-5) 276 1380, Email: info@teltonika. 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

Project Networks With Mixed-Time Constraints

Project Networks With Mixed-Time Constraints Project Networs Wth Mxed-Tme Constrants L Caccetta and B Wattananon Western Australan Centre of Excellence n Industral Optmsaton (WACEIO) Curtn Unversty of Technology GPO Box U1987 Perth Western Australa

More information

QoS-Aware Active Queue Management for Multimedia Services over the Internet

QoS-Aware Active Queue Management for Multimedia Services over the Internet QoS-Aware Actve Queue Management for Multmeda Servces over the Internet I-Shyan Hwang, *Bor-Junn Hwang, Pen-Mng Chang, Cheng-Yu Wang Abstract Recently, the multmeda servces such as IPTV, vdeo conference

More information

A Replication-Based and Fault Tolerant Allocation Algorithm for Cloud Computing

A Replication-Based and Fault Tolerant Allocation Algorithm for Cloud Computing A Replcaton-Based and Fault Tolerant Allocaton Algorthm for Cloud Computng Tork Altameem Dept of Computer Scence, RCC, Kng Saud Unversty, PO Box: 28095 11437 Ryadh-Saud Araba Abstract The very large nfrastructure

More information

INVESTIGATION OF VEHICULAR USERS FAIRNESS IN CDMA-HDR NETWORKS

INVESTIGATION OF VEHICULAR USERS FAIRNESS IN CDMA-HDR NETWORKS 21 22 September 2007, BULGARIA 119 Proceedngs of the Internatonal Conference on Informaton Technologes (InfoTech-2007) 21 st 22 nd September 2007, Bulgara vol. 2 INVESTIGATION OF VEHICULAR USERS FAIRNESS

More information

CHOLESTEROL REFERENCE METHOD LABORATORY NETWORK. Sample Stability Protocol

CHOLESTEROL REFERENCE METHOD LABORATORY NETWORK. Sample Stability Protocol CHOLESTEROL REFERENCE METHOD LABORATORY NETWORK Sample Stablty Protocol Background The Cholesterol Reference Method Laboratory Network (CRMLN) developed certfcaton protocols for total cholesterol, HDL

More information

Minimal Coding Network With Combinatorial Structure For Instantaneous Recovery From Edge Failures

Minimal Coding Network With Combinatorial Structure For Instantaneous Recovery From Edge Failures Mnmal Codng Network Wth Combnatoral Structure For Instantaneous Recovery From Edge Falures Ashly Joseph 1, Mr.M.Sadsh Sendl 2, Dr.S.Karthk 3 1 Fnal Year ME CSE Student Department of Computer Scence Engneerng

More information

M3S MULTIMEDIA MOBILITY MANAGEMENT AND LOAD BALANCING IN WIRELESS BROADCAST NETWORKS

M3S MULTIMEDIA MOBILITY MANAGEMENT AND LOAD BALANCING IN WIRELESS BROADCAST NETWORKS M3S MULTIMEDIA MOBILITY MANAGEMENT AND LOAD BALANCING IN WIRELESS BROADCAST NETWORKS Bogdan Cubotaru, Gabrel-Mro Muntean Performance Engneerng Laboratory, RINCE School of Electronc Engneerng Dubln Cty

More information

[email protected]@cityu.edu.hk [email protected], [email protected]

taposh_kuet20@yahoo.comcsedchan@cityu.edu.hk rajib_csedept@yahoo.co.uk, alam_shihabul@yahoo.com G. G. Md. Nawaz Al 1,2, Rajb Chakraborty 2, Md. Shhabul Alam 2 and Edward Chan 1 1 Cty Unversty of Hong Kong, Hong Kong, Chna [email protected]@ctyu.edu.hk 2 Khulna Unversty of Engneerng

More information

Reinforcement Learning for Quality of Service in Mobile Ad Hoc Network (MANET)

Reinforcement Learning for Quality of Service in Mobile Ad Hoc Network (MANET) Renforcement Learnng for Qualty of Servce n Moble Ad Hoc Network (MANET) *T.KUMANAN AND **K.DURAISWAMY *Meenaksh College of Engneerng West K.K Nagar, Cheena-78 **Dean/academc,K.S.R College of Technology,Truchengode

More information

APPLICATION OF PROBE DATA COLLECTED VIA INFRARED BEACONS TO TRAFFIC MANEGEMENT

APPLICATION OF PROBE DATA COLLECTED VIA INFRARED BEACONS TO TRAFFIC MANEGEMENT APPLICATION OF PROBE DATA COLLECTED VIA INFRARED BEACONS TO TRAFFIC MANEGEMENT Toshhko Oda (1), Kochro Iwaoka (2) (1), (2) Infrastructure Systems Busness Unt, Panasonc System Networks Co., Ltd. Saedo-cho

More information

RESEARCH ON DUAL-SHAKER SINE VIBRATION CONTROL. Yaoqi FENG 1, Hanping QIU 1. China Academy of Space Technology (CAST) yaoqi.feng@yahoo.

RESEARCH ON DUAL-SHAKER SINE VIBRATION CONTROL. Yaoqi FENG 1, Hanping QIU 1. China Academy of Space Technology (CAST) yaoqi.feng@yahoo. ICSV4 Carns Australa 9- July, 007 RESEARCH ON DUAL-SHAKER SINE VIBRATION CONTROL Yaoq FENG, Hanpng QIU Dynamc Test Laboratory, BISEE Chna Academy of Space Technology (CAST) [email protected] Abstract

More information

denote the location of a node, and suppose node X . This transmission causes a successful reception by node X for any other node

denote the location of a node, and suppose node X . This transmission causes a successful reception by node X for any other node Fnal Report of EE359 Class Proect Throughput and Delay n Wreless Ad Hoc Networs Changhua He [email protected] Abstract: Networ throughput and pacet delay are the two most mportant parameters to evaluate

More information

Feature selection for intrusion detection. Slobodan Petrović NISlab, Gjøvik University College

Feature selection for intrusion detection. Slobodan Petrović NISlab, Gjøvik University College Feature selecton for ntruson detecton Slobodan Petrovć NISlab, Gjøvk Unversty College Contents The feature selecton problem Intruson detecton Traffc features relevant for IDS The CFS measure The mrmr measure

More information

QoS in the Linux Operating System. Technical Report

QoS in the Linux Operating System. Technical Report Unverstät Karlsruhe (H) Insttut für elematk QoS n the Lnux Operatng System echncal Report Marc Bechler and Hartmut Rtter Insttut für elematk Fakultät für Informatk Unverstät Karlsruhe (H) E-Mal: [mbechler

More information

Performance Analysis and Comparison of QoS Provisioning Mechanisms for CBR Traffic in Noisy IEEE 802.11e WLANs Environments

Performance Analysis and Comparison of QoS Provisioning Mechanisms for CBR Traffic in Noisy IEEE 802.11e WLANs Environments Tamkang Journal of Scence and Engneerng, Vol. 12, No. 2, pp. 143149 (2008) 143 Performance Analyss and Comparson of QoS Provsonng Mechansms for CBR Traffc n Nosy IEEE 802.11e WLANs Envronments Der-Junn

More information

Fair Virtual Bandwidth Allocation Model in Virtual Data Centers

Fair Virtual Bandwidth Allocation Model in Virtual Data Centers Far Vrtual Bandwdth Allocaton Model n Vrtual Data Centers Yng Yuan, Cu-rong Wang, Cong Wang School of Informaton Scence and Engneerng ortheastern Unversty Shenyang, Chna School of Computer and Communcaton

More information

ANALYZING THE RELATIONSHIPS BETWEEN QUALITY, TIME, AND COST IN PROJECT MANAGEMENT DECISION MAKING

ANALYZING THE RELATIONSHIPS BETWEEN QUALITY, TIME, AND COST IN PROJECT MANAGEMENT DECISION MAKING ANALYZING THE RELATIONSHIPS BETWEEN QUALITY, TIME, AND COST IN PROJECT MANAGEMENT DECISION MAKING Matthew J. Lberatore, Department of Management and Operatons, Vllanova Unversty, Vllanova, PA 19085, 610-519-4390,

More information

Luby s Alg. for Maximal Independent Sets using Pairwise Independence

Luby s Alg. for Maximal Independent Sets using Pairwise Independence Lecture Notes for Randomzed Algorthms Luby s Alg. for Maxmal Independent Sets usng Parwse Independence Last Updated by Erc Vgoda on February, 006 8. Maxmal Independent Sets For a graph G = (V, E), an ndependent

More information

Course outline. Financial Time Series Analysis. Overview. Data analysis. Predictive signal. Trading strategy

Course outline. Financial Time Series Analysis. Overview. Data analysis. Predictive signal. Trading strategy Fnancal Tme Seres Analyss Patrck McSharry [email protected] www.mcsharry.net Trnty Term 2014 Mathematcal Insttute Unversty of Oxford Course outlne 1. Data analyss, probablty, correlatons, vsualsaton

More information

iavenue iavenue i i i iavenue iavenue iavenue

iavenue iavenue i i i iavenue iavenue iavenue Saratoga Systems' enterprse-wde Avenue CRM system s a comprehensve web-enabled software soluton. Ths next generaton system enables you to effectvely manage and enhance your customer relatonshps n both

More information

DBA-VM: Dynamic Bandwidth Allocator for Virtual Machines

DBA-VM: Dynamic Bandwidth Allocator for Virtual Machines DBA-VM: Dynamc Bandwdth Allocator for Vrtual Machnes Ahmed Amamou, Manel Bourguba, Kamel Haddadou and Guy Pujolle LIP6, Perre & Mare Cure Unversty, 4 Place Jusseu 755 Pars, France Gand SAS, 65 Boulevard

More information

Module 2 LOSSLESS IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur

Module 2 LOSSLESS IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur Module LOSSLESS IMAGE COMPRESSION SYSTEMS Lesson 3 Lossless Compresson: Huffman Codng Instructonal Objectves At the end of ths lesson, the students should be able to:. Defne and measure source entropy..

More information

A Design Method of High-availability and Low-optical-loss Optical Aggregation Network Architecture

A Design Method of High-availability and Low-optical-loss Optical Aggregation Network Architecture A Desgn Method of Hgh-avalablty and Low-optcal-loss Optcal Aggregaton Network Archtecture Takehro Sato, Kuntaka Ashzawa, Kazumasa Tokuhash, Dasuke Ish, Satoru Okamoto and Naoak Yamanaka Dept. of Informaton

More information

An agent architecture for network support of distributed simulation systems

An agent architecture for network support of distributed simulation systems An agent archtecture for network support of dstrbuted smulaton systems Robert Smon, Mark Pullen and Woan Sun Chang Department of Computer Scence George Mason Unversty Farfax, VA, 22032 U.S.A. smon, mpullen,

More information

A Cluster Based Replication Architecture for Load Balancing in Peer-to-Peer Content Distribution

A Cluster Based Replication Architecture for Load Balancing in Peer-to-Peer Content Distribution A Cluster Based Replcaton Archtecture for Load Balancng n Peer-to-Peer Content Dstrbuton S.Ayyasamy 1 and S.N. Svanandam 2 1 Asst. Professor, Department of Informaton Technology, Tamlnadu College of Engneerng

More information

Multi-Source Video Multicast in Peer-to-Peer Networks

Multi-Source Video Multicast in Peer-to-Peer Networks ult-source Vdeo ultcast n Peer-to-Peer Networks Francsco de Asís López-Fuentes*, Eckehard Stenbach Technsche Unverstät ünchen Insttute of Communcaton Networks, eda Technology Group 80333 ünchen, Germany

More information

Cooperative Load Balancing in IEEE 802.11 Networks with Cell Breathing

Cooperative Load Balancing in IEEE 802.11 Networks with Cell Breathing Cooperatve Load Balancng n IEEE 82.11 Networks wth Cell Breathng Eduard Garca Rafael Vdal Josep Paradells Wreless Networks Group - Techncal Unversty of Catalona (UPC) {eduardg, rvdal, teljpa}@entel.upc.edu;

More information

A Parallel Architecture for Stateful Intrusion Detection in High Traffic Networks

A Parallel Architecture for Stateful Intrusion Detection in High Traffic Networks A Parallel Archtecture for Stateful Intruson Detecton n Hgh Traffc Networks Mchele Colajann Mrco Marchett Dpartmento d Ingegnera dell Informazone Unversty of Modena {colajann, marchett.mrco}@unmore.t Abstract

More information

Network Aware Load-Balancing via Parallel VM Migration for Data Centers

Network Aware Load-Balancing via Parallel VM Migration for Data Centers Network Aware Load-Balancng va Parallel VM Mgraton for Data Centers Kun-Tng Chen 2, Chen Chen 12, Po-Hsang Wang 2 1 Informaton Technology Servce Center, 2 Department of Computer Scence Natonal Chao Tung

More information

Calculating the high frequency transmission line parameters of power cables

Calculating the high frequency transmission line parameters of power cables < ' Calculatng the hgh frequency transmsson lne parameters of power cables Authors: Dr. John Dcknson, Laboratory Servces Manager, N 0 RW E B Communcatons Mr. Peter J. Ncholson, Project Assgnment Manager,

More information

On the Optimal Control of a Cascade of Hydro-Electric Power Stations

On the Optimal Control of a Cascade of Hydro-Electric Power Stations On the Optmal Control of a Cascade of Hydro-Electrc Power Statons M.C.M. Guedes a, A.F. Rbero a, G.V. Smrnov b and S. Vlela c a Department of Mathematcs, School of Scences, Unversty of Porto, Portugal;

More information

How To Solve An Onlne Control Polcy On A Vrtualzed Data Center

How To Solve An Onlne Control Polcy On A Vrtualzed Data Center Dynamc Resource Allocaton and Power Management n Vrtualzed Data Centers Rahul Urgaonkar, Ulas C. Kozat, Ken Igarash, Mchael J. Neely [email protected], {kozat, garash}@docomolabs-usa.com, [email protected]

More information

A Passive Network Measurement-based Traffic Control Algorithm in Gateway of. P2P Systems

A Passive Network Measurement-based Traffic Control Algorithm in Gateway of. P2P Systems roceedngs of the 7th World Congress The Internatonal Federaton of Automatc Control A assve Network Measurement-based Traffc Control Algorthm n Gateway of 2 Systems Ybo Jang, Weje Chen, Janwe Zheng, Wanlang

More information

What is Candidate Sampling

What is Candidate Sampling What s Canddate Samplng Say we have a multclass or mult label problem where each tranng example ( x, T ) conssts of a context x a small (mult)set of target classes T out of a large unverse L of possble

More information

An Alternative Way to Measure Private Equity Performance

An Alternative Way to Measure Private Equity Performance An Alternatve Way to Measure Prvate Equty Performance Peter Todd Parlux Investment Technology LLC Summary Internal Rate of Return (IRR) s probably the most common way to measure the performance of prvate

More information

Application of Multi-Agents for Fault Detection and Reconfiguration of Power Distribution Systems

Application of Multi-Agents for Fault Detection and Reconfiguration of Power Distribution Systems 1 Applcaton of Mult-Agents for Fault Detecton and Reconfguraton of Power Dstrbuton Systems K. Nareshkumar, Member, IEEE, M. A. Choudhry, Senor Member, IEEE, J. La, A. Felach, Senor Member, IEEE Abstract--The

More information

Network Security Situation Evaluation Method for Distributed Denial of Service

Network Security Situation Evaluation Method for Distributed Denial of Service Network Securty Stuaton Evaluaton Method for Dstrbuted Denal of Servce Jn Q,2, Cu YMn,2, Huang MnHuan,2, Kuang XaoHu,2, TangHong,2 ) Scence and Technology on Informaton System Securty Laboratory, Bejng,

More information

Traffic-light a stress test for life insurance provisions

Traffic-light a stress test for life insurance provisions MEMORANDUM Date 006-09-7 Authors Bengt von Bahr, Göran Ronge Traffc-lght a stress test for lfe nsurance provsons Fnansnspetonen P.O. Box 6750 SE-113 85 Stocholm [Sveavägen 167] Tel +46 8 787 80 00 Fax

More information

Cloud Auto-Scaling with Deadline and Budget Constraints

Cloud Auto-Scaling with Deadline and Budget Constraints Prelmnary verson. Fnal verson appears In Proceedngs of 11th ACM/IEEE Internatonal Conference on Grd Computng (Grd 21). Oct 25-28, 21. Brussels, Belgum. Cloud Auto-Scalng wth Deadlne and Budget Constrants

More information

A Programming Model for the Cloud Platform

A Programming Model for the Cloud Platform Internatonal Journal of Advanced Scence and Technology A Programmng Model for the Cloud Platform Xaodong Lu School of Computer Engneerng and Scence Shangha Unversty, Shangha 200072, Chna [email protected]

More information

Frequency Selective IQ Phase and IQ Amplitude Imbalance Adjustments for OFDM Direct Conversion Transmitters

Frequency Selective IQ Phase and IQ Amplitude Imbalance Adjustments for OFDM Direct Conversion Transmitters Frequency Selectve IQ Phase and IQ Ampltude Imbalance Adjustments for OFDM Drect Converson ransmtters Edmund Coersmeer, Ernst Zelnsk Noka, Meesmannstrasse 103, 44807 Bochum, Germany [email protected],

More information

Network Design and Appraisal

Network Design and Appraisal Qualty-of-Servce IP QoS archtectures Ulrch Fedler ([email protected]) Introductory Example Applcaton: IP telephony wth fxed encodng bt rate Network delay s varable, packets may get lost Network qualty-of

More information

A Load-Balancing Algorithm for Cluster-based Multi-core Web Servers

A Load-Balancing Algorithm for Cluster-based Multi-core Web Servers Journal of Computatonal Informaton Systems 7: 13 (2011) 4740-4747 Avalable at http://www.jofcs.com A Load-Balancng Algorthm for Cluster-based Mult-core Web Servers Guohua YOU, Yng ZHAO College of Informaton

More information

On File Delay Minimization for Content Uploading to Media Cloud via Collaborative Wireless Network

On File Delay Minimization for Content Uploading to Media Cloud via Collaborative Wireless Network On Fle Delay Mnmzaton for Content Uploadng to Meda Cloud va Collaboratve Wreless Network Ge Zhang and Yonggang Wen School of Computer Engneerng Nanyang Technologcal Unversty Sngapore Emal: {zh0001ge, ygwen}@ntu.edu.sg

More information

Forecasting the Demand of Emergency Supplies: Based on the CBR Theory and BP Neural Network

Forecasting the Demand of Emergency Supplies: Based on the CBR Theory and BP Neural Network 700 Proceedngs of the 8th Internatonal Conference on Innovaton & Management Forecastng the Demand of Emergency Supples: Based on the CBR Theory and BP Neural Network Fu Deqang, Lu Yun, L Changbng School

More information

A Distributed Algorithm for Least Constraining Slot Allocation in MPLS Optical TDM Networks

A Distributed Algorithm for Least Constraining Slot Allocation in MPLS Optical TDM Networks A Dstrbuted Algorthm for Least Constranng Slot Allocaton n MPLS Optcal TDM Networks Hassan Zeneddne and Gregor V. Bochmann, Unversty of Ottawa, Ottawa, ON-K1N 6N5 Abstract - In ths paper, we propose a

More information

IMPACT ANALYSIS OF A CELLULAR PHONE

IMPACT ANALYSIS OF A CELLULAR PHONE 4 th ASA & μeta Internatonal Conference IMPACT AALYSIS OF A CELLULAR PHOE We Lu, 2 Hongy L Bejng FEAonlne Engneerng Co.,Ltd. Bejng, Chna ABSTRACT Drop test smulaton plays an mportant role n nvestgatng

More information

ivoip: an Intelligent Bandwidth Management Scheme for VoIP in WLANs

ivoip: an Intelligent Bandwidth Management Scheme for VoIP in WLANs VoIP: an Intellgent Bandwdth Management Scheme for VoIP n WLANs Zhenhu Yuan and Gabrel-Mro Muntean Abstract Voce over Internet Protocol (VoIP) has been wdely used by many moble consumer devces n IEEE 802.11

More information

To manage leave, meeting institutional requirements and treating individual staff members fairly and consistently.

To manage leave, meeting institutional requirements and treating individual staff members fairly and consistently. Corporate Polces & Procedures Human Resources - Document CPP216 Leave Management Frst Produced: Current Verson: Past Revsons: Revew Cycle: Apples From: 09/09/09 26/10/12 09/09/09 3 years Immedately Authorsaton:

More information

P2P/ Grid-based Overlay Architecture to Support VoIP Services in Large Scale IP Networks

P2P/ Grid-based Overlay Architecture to Support VoIP Services in Large Scale IP Networks PP/ Grd-based Overlay Archtecture to Support VoIP Servces n Large Scale IP Networks We Yu *, Srram Chellappan # and Dong Xuan # * Dept. of Computer Scence, Texas A&M Unversty, U.S.A. {weyu}@cs.tamu.edu

More information

A Novel Adaptive Load Balancing Routing Algorithm in Ad hoc Networks

A Novel Adaptive Load Balancing Routing Algorithm in Ad hoc Networks Journal of Convergence Informaton Technology A Novel Adaptve Load Balancng Routng Algorthm n Ad hoc Networks Zhu Bn, Zeng Xao-png, Xong Xan-sheng, Chen Qan, Fan Wen-yan, We Geng College of Communcaton

More information

Data Broadcast on a Multi-System Heterogeneous Overlayed Wireless Network *

Data Broadcast on a Multi-System Heterogeneous Overlayed Wireless Network * JOURNAL OF INFORMATION SCIENCE AND ENGINEERING 24, 819-840 (2008) Data Broadcast on a Mult-System Heterogeneous Overlayed Wreless Network * Department of Computer Scence Natonal Chao Tung Unversty Hsnchu,

More information

How To Plan A Network Wide Load Balancing Route For A Network Wde Network (Network)

How To Plan A Network Wide Load Balancing Route For A Network Wde Network (Network) Network-Wde Load Balancng Routng Wth Performance Guarantees Kartk Gopalan Tz-cker Chueh Yow-Jan Ln Florda State Unversty Stony Brook Unversty Telcorda Research [email protected] [email protected] [email protected]

More information

RequIn, a tool for fast web traffic inference

RequIn, a tool for fast web traffic inference RequIn, a tool for fast web traffc nference Olver aul, Jean Etenne Kba GET/INT, LOR Department 9 rue Charles Fourer 90 Evry, France [email protected], [email protected] Abstract As networked

More information

A New Task Scheduling Algorithm Based on Improved Genetic Algorithm

A New Task Scheduling Algorithm Based on Improved Genetic Algorithm A New Task Schedulng Algorthm Based on Improved Genetc Algorthm n Cloud Computng Envronment Congcong Xong, Long Feng, Lxan Chen A New Task Schedulng Algorthm Based on Improved Genetc Algorthm n Cloud Computng

More information

A Dynamic Load Balancing for Massive Multiplayer Online Game Server

A Dynamic Load Balancing for Massive Multiplayer Online Game Server A Dynamc Load Balancng for Massve Multplayer Onlne Game Server Jungyoul Lm, Jaeyong Chung, Jnryong Km and Kwanghyun Shm Dgtal Content Research Dvson Electroncs and Telecommuncatons Research Insttute Daejeon,

More information

An Inter-Frame De-Jittering Scheme for Video Streaming over Mobile Communication Networks

An Inter-Frame De-Jittering Scheme for Video Streaming over Mobile Communication Networks An Inter-Frame De-Jtterng Scheme for Vdeo Streamng over Moble Communcaton Networks Tsang-Lng Sheu and Po-Wen Lee Department of Electrcal Engneerng Natonal Sun Yat-Sen Unversty Kaohsung, Tawan [email protected]

More information

Enterprise Master Patient Index

Enterprise Master Patient Index Enterprse Master Patent Index Healthcare data are captured n many dfferent settngs such as hosptals, clncs, labs, and physcan offces. Accordng to a report by the CDC, patents n the Unted States made an

More information

Hosting Virtual Machines on Distributed Datacenters

Hosting Virtual Machines on Distributed Datacenters Hostng Vrtual Machnes on Dstrbuted Datacenters Chuan Pham Scence and Engneerng, KyungHee Unversty, Korea [email protected] Jae Hyeok Son Scence and Engneerng, KyungHee Unversty, Korea [email protected]

More information

How To Understand The Results Of The German Meris Cloud And Water Vapour Product

How To Understand The Results Of The German Meris Cloud And Water Vapour Product Ttel: Project: Doc. No.: MERIS level 3 cloud and water vapour products MAPP MAPP-ATBD-ClWVL3 Issue: 1 Revson: 0 Date: 9.12.1998 Functon Name Organsaton Sgnature Date Author: Bennartz FUB Preusker FUB Schüller

More information

A GENERIC HANDOVER DECISION MANAGEMENT FRAMEWORK FOR NEXT GENERATION NETWORKS

A GENERIC HANDOVER DECISION MANAGEMENT FRAMEWORK FOR NEXT GENERATION NETWORKS A GENERIC HANDOVER DECISION MANAGEMENT FRAMEWORK FOR NEXT GENERATION NETWORKS Shanthy Menezes 1 and S. Venkatesan 2 1 Department of Computer Scence, Unversty of Texas at Dallas, Rchardson, TX, USA 1 [email protected]

More information

1. Fundamentals of probability theory 2. Emergence of communication traffic 3. Stochastic & Markovian Processes (SP & MP)

1. Fundamentals of probability theory 2. Emergence of communication traffic 3. Stochastic & Markovian Processes (SP & MP) 6.3 / -- Communcaton Networks II (Görg) SS20 -- www.comnets.un-bremen.de Communcaton Networks II Contents. Fundamentals of probablty theory 2. Emergence of communcaton traffc 3. Stochastc & Markovan Processes

More information

Open Access A Load Balancing Strategy with Bandwidth Constraint in Cloud Computing. Jing Deng 1,*, Ping Guo 2, Qi Li 3, Haizhu Chen 1

Open Access A Load Balancing Strategy with Bandwidth Constraint in Cloud Computing. Jing Deng 1,*, Ping Guo 2, Qi Li 3, Haizhu Chen 1 Send Orders for Reprnts to [email protected] The Open Cybernetcs & Systemcs Journal, 2014, 8, 115-121 115 Open Access A Load Balancng Strategy wth Bandwdth Constrant n Cloud Computng Jng Deng 1,*,

More information

Canon NTSC Help Desk Documentation

Canon NTSC Help Desk Documentation Canon NTSC Help Desk Documentaton READ THIS BEFORE PROCEEDING Before revewng ths documentaton, Canon Busness Solutons, Inc. ( CBS ) hereby refers you, the customer or customer s representatve or agent

More information

Robust Design of Public Storage Warehouses. Yeming (Yale) Gong EMLYON Business School

Robust Design of Public Storage Warehouses. Yeming (Yale) Gong EMLYON Business School Robust Desgn of Publc Storage Warehouses Yemng (Yale) Gong EMLYON Busness School Rene de Koster Rotterdam school of management, Erasmus Unversty Abstract We apply robust optmzaton and revenue management

More information

DEFINING %COMPLETE IN MICROSOFT PROJECT

DEFINING %COMPLETE IN MICROSOFT PROJECT CelersSystems DEFINING %COMPLETE IN MICROSOFT PROJECT PREPARED BY James E Aksel, PMP, PMI-SP, MVP For Addtonal Informaton about Earned Value Management Systems and reportng, please contact: CelersSystems,

More information

Peer-to-Peer Networks Protocols, Cooperation and Competition

Peer-to-Peer Networks Protocols, Cooperation and Competition Peer-to-Peer Networks Protocols, Cooperaton and Competton Hyunggon Park Sgnal Processng Laboratory (LTS4), Insttute of Electrcal Engneerng, Swss Federal Insttute of Technology (EPFL), Lausanne, Swtzerland

More information

Enabling P2P One-view Multi-party Video Conferencing

Enabling P2P One-view Multi-party Video Conferencing Enablng P2P One-vew Mult-party Vdeo Conferencng Yongxang Zhao, Yong Lu, Changja Chen, and JanYn Zhang Abstract Mult-Party Vdeo Conferencng (MPVC) facltates realtme group nteracton between users. Whle P2P

More information

An Interest-Oriented Network Evolution Mechanism for Online Communities

An Interest-Oriented Network Evolution Mechanism for Online Communities An Interest-Orented Network Evoluton Mechansm for Onlne Communtes Cahong Sun and Xaopng Yang School of Informaton, Renmn Unversty of Chna, Bejng 100872, P.R. Chna {chsun,yang}@ruc.edu.cn Abstract. Onlne

More information

Self-Adaptive SLA-Driven Capacity Management for Internet Services

Self-Adaptive SLA-Driven Capacity Management for Internet Services Self-Adaptve SLA-Drven Capacty Management for Internet Servces Bruno Abrahao, Vrglo Almeda and Jussara Almeda Computer Scence Department Federal Unversty of Mnas Geras, Brazl Alex Zhang, Drk Beyer and

More information

The OC Curve of Attribute Acceptance Plans

The OC Curve of Attribute Acceptance Plans The OC Curve of Attrbute Acceptance Plans The Operatng Characterstc (OC) curve descrbes the probablty of acceptng a lot as a functon of the lot s qualty. Fgure 1 shows a typcal OC Curve. 10 8 6 4 1 3 4

More information

FORMAL ANALYSIS FOR REAL-TIME SCHEDULING

FORMAL ANALYSIS FOR REAL-TIME SCHEDULING FORMAL ANALYSIS FOR REAL-TIME SCHEDULING Bruno Dutertre and Vctora Stavrdou, SRI Internatonal, Menlo Park, CA Introducton In modern avoncs archtectures, applcaton software ncreasngly reles on servces provded

More information