Chapter 2 Valiant Load-Balancing: Building Networks That Can Support All Traffic Matrices

Size: px
Start display at page:

Download "Chapter 2 Valiant Load-Balancing: Building Networks That Can Support All Traffic Matrices"

Transcription

1 Chapte 2 Valiant Load-Balancing: Building etwoks That Can Suppot All Taffic Matices Rui Zhang-Shen Abstact This pape is a bief suvey on how Valiant load-balancing (VLB) can be used to build netwoks that can efficiently and eliably suppot all taffic matices. We discuss how to extend VLB to netwoks with heteogeneous capacities, how to potect against failues in a VLB netwok, and how to inteconnect two VLB netwoks. Fo the eades efeence, included also is a list of wok that uses VLB in vaious aspects of netwoking. 2.1 Intoduction In many netwoks the taffic matix is eithe had to measue and pedict, o highly vaiable ove time. In these cases, using Valiant load-balancing (VLB) to suppot all possible taffic matices is an attactive option. Fo example, even though the taffic in the Intenet backbone is extemely smooth due to high level of aggegation, it is still had to measue. Accuately measuing the taffic matix (e.g., using etflow) is too expensive to do all the time, and standad methods using link measuements give eos of 20% o moe. Even if the cuent taffic matix is satisfactoily obtained, extapolating it to the futue is faught with uncetainty, due to the unpedictable natue of Intenet taffic gowth. Finally, since Intenet taffic is dynamic, the taffic matix can deviate fom its nomal values at any time, possibly causing congestion. The taffic demand seen by a netwok can be epesented by the taffic matix, which indicates the ates at which each node initiates taffic to evey othe node. We say a netwok can suppot a taffic matix if fo evey link in the netwok, the load caused by the taffic matix is less than the capacity of the link. When a netwok cannot suppot the taffic matix pesented to it, at least one link in the netwok has a load highe than its capacity. Congestion occus and backlog in the buffe builds up on the congested link(s), causing packet dops, inceased delay, and high vaiations R. Zhang-Shen ( ) Google, Inc., ew Yok, Y uizhangshen@gmail.com G. Comode and M. Thottan (eds.), Algoithms fo ext Geneation etwoks, Compute Communications and etwoks, DOI / , c Spinge-Velag London Limited

2 20 R. Zhang-Shen in delay. Ideally, we would like to design a netwok that can suppot a wide ange of taffic matices, so that congestion occus only aely o not at all. In this pape, we discuss the use of VLB in building netwoks that can efficiently suppot all taffic matices which do not ove-subscibe any node. We fist biefly suvey the wide use of VLB in vaious aspects of netwoking, and descibe the basic scenaio of using VLB in a netwok. Section 2.2 extends VLB fom a homogeneous setting to netwoks with abitay capacities, Section 2.3 descibes how to potect against and ecove quickly fom failues in a VLB netwok, and Section 2.4 poposes to use VLB to oute taffic between two netwoks. Finally Section 2.5 discusses possible futue wok The Wide Use of VLB In the ealy 1980s, Valiant [19] fist poposed the scheme of outing though a andomly picked intemediate node en oute to a packet s destination. He showed that in an -node binay cube netwok, given any pemutation taffic matix, the distibuted two-phase andomized outing can oute evey packet to its destination within O(log ) time with ovewhelming pobability. This was the fist scheme fo outing abitay pemutation in a spase netwok in O(log ) time. Since then, such andomized outing has been used widely, and is often efeed to as (VLB), andomized load-balancing,otwo-phase outing. VLB has many good chaacteistics. It is decentalized, whee evey node makes local decisions. This also makes the scheme scalable. VLB is agnostic to the taffic matix because the andomness eases the taffic patten, and diffeent taffic matices can esult in the same load on the links. Soon afte its invention, was used in othe inteconnection netwoks fo paallel communication, to impove delivey time [1], and to elieve effects of advese taffic pattens [13]. In ecent yeas, it was adapted fo outing in tous netwoks [17, 18] in ode to povide wost-case pefomance guaantees without sacificing aveagecase pefomance. The key is to use VLB adaptively, based on the obsevation that unde low load, load-balancing only a small amount of taffic is sufficient to avoid congestion. VLB is also used in building netwok switches with geat scalability and pefomance guaantee, without the need of a centalized schedule. It was used in ATM switches [7], outes [4, 5], opticaloutes [3, 9], andsoftwaeoutes [2]. Inpaticula,theschemewas ediscoveedfodesigningouteswitch fabics [4] to mitigate outes scaling challenges, because it was difficult fo centalized schemes to keep up with the inceasing link speed. In this context, it was shown that splitting taffic in a ound-obin fashion has the same effect on link load as andom splitting [4], and that is the most efficient in tems of the total equied inteconnection capacity fo suppoting all taffic matices [8]. Almost simultaneously, seveal goups independently applied the idea of VLB to taffic engineeing and netwok design fo the Intenet, in ode to efficiently sup-

3 2 Valiant Load-Balancing: Building etwoks That Can Suppot All Taffic Matices 21 pot all possible taffic matices. Kodialam et al. s two-phase outing [11, 12] isa taffic engineeing method, whee a full mesh of tunnels ae set up ove fixed capacity links and packets ae sent in two phases (i.e., two hops) in the netwok. Winze et al. s selective andomized load-balancing [14, 16, 21] used VLB and its vaiants to design cost-effective optical netwoks. Thei model assumes that a link s cost includes both the fibe and the teminating equipment, so thee is incentive fo having fewe links. In the optimal design, taffic is load-balanced only to a few intemediate nodes. Zhang-Shen and McKewon [23, 25] poposed using VLB ove a logical full mesh in a backbone netwok to suppot all taffic matices and to quickly ecove fom failues. In addition, VLB was used as an optical outing stategy in Ethenet LA [20], fo scheduling in meto aea WDM ings [10], fo cicuit-switched netwoks [22], and fo scaling and commoditizing data cente netwoks [6]. A study on the queueing popeties of a VLB netwok [15] found that VLB eliminates congestion in the netwok, and pseudo-andom (e.g., ound-obin) loadbalancing educes queueing delay. VLB was also shown to eliminate congestion on peeing links when used to oute taffic between netwoks [26] A Simple VLB etwok Conside a netwok of nodes, each with capacity, i.e., a node can initiate taffic at the maximum ate of, and can eceive taffic at the same maximum ate. We assume that the netwok taffic satisfies such node aggegate constaint, because othewise thee is no way to avoid congestion. A logical link of capacity 2 is established between evey pai of nodes ove the physical links, as shown in Figue 2.1. We use the convention that a flow in the netwok is defined by the souce node and the destination node, unless futhe specified. Evey flow enteing the netwok is equally split acoss two-hop paths between ingess and egess nodes, i.e., a packet is fowaded twice in the netwok: In the fist hop, an ingess node unifomly distibutes each of its incoming flows to all the nodes, egadless of 1 2/ 2 3 Fig. 2.1 VLB in a netwok of identical nodes each having capacity.afull mesh of logical links of capacity 2= connect the nodes... 4

4 22 R. Zhang-Shen the destinations. In the second hop, all packets ae sent to the final destinations by the intemediate nodes. Load-balancing can be done packet-by-packet, o flow-byflow at the application flow level. The splitting of taffic can be andom (e.g., to a andomly picked intemediate node) o deteministic (e.g., ound-obin). Assume we can achieve pefect load-balancing, i.e., can split taffic at the exact popotions we desie, then each node eceives exactly 1 of evey flow afte fist-hop outing. This means, all the nodes equally shae the buden of fowading taffic as the intemediate node. When the intemediate node happens to be the ingess o egess node, the flow actually taveses one hop (the diect link 2 between ingess and egess) in the netwok. Hence, of evey flow taveses the coesponding one-hop path. Such unifom load-balancing can guaantee to suppot all taffic matices in this netwok. Since the incoming taffic ate to each node is at most, andthetafficis evenly load-balanced to nodes, the actual taffic on each link due to the fist-hop outing is at most. The second-hop outing is the dual of the fist-hop outing. Since each node can eceive taffic at a maximum ate of and eceives 1 of the taffic fom evey node, the actual taffic on each link due to the second-hop outing is also at most 2. Theefoe, a full-mesh netwok whee each link has capacity is sufficient to suppot all taffic matices in a netwok of nodes of capacity. This is pehaps a supising esult a netwok whee any two nodes ae connected with a link of capacity 2 can suppot taffic matices whee a node can send taffic to anothe node at ate. It shows the powe of load-balancing. In VLB, each flow is caied by paths, and each link caies a faction of many flows; theefoe any lage flow is aveaged out by othe small flows. In a static full-mesh netwok, if all the taffic wee to be sent though diect paths, we would need a full-mesh netwok of link capacity to suppot all possible taffic matices; theefoe load-balancing is 2 times moe efficient than diect outing. 2.2 VLB in Heteogeneous etwoks Real-life netwoks ae often heteogeneous, i.e., the nodes in a netwok can have diffeent capacities. In this section we discuss how to extend the esult of unifom load-balancing to heteogeneous netwoks [24]. We fist intoduce notations. In a netwok of nodes, the taffic matix D f ij g is an matix, whee the enty ij indicates the dataate at which ode i initiates taffic destined to ode j. The taffic ate is typically aveaged ove a long peiod of time so we conside it as constant. Typically in a netwok thee ae buffes to absob shot-lived taffic fluctuations. Suppose ode i has capacity i, i.e., the node can initiate taffic at the maximum ate of i, and can eceive taffic at the same maximum ate. So in ode fo the taffic matix to not ove-subscibe any node, it must satisfy X X ij i ; 8i and ji i ; 8i: (2.1) j j

5 2 Valiant Load-Balancing: Building etwoks That Can Suppot All Taffic Matices 23 Without loss of geneality, assume that the nodes have been soted accoding to deceasing capacities, i.e., 1 2, so ode 1 is the lagest node and ode the smallest. Let R be the total node capacity, i.e., R D P id1 i.we assume 1 P id2 i because even if 1 > P id2 i, ode 1 cannot send o eceive taffic at a ate highe than P id2 i, because that would ove-subscibe some nodes. Suppose that a full mesh of logical links ae set up to connect these nodes. Let c ij epesent the equied link capacity fom ode i to ode j and C the link capacity matix fc ij g.havingc ij D 0 means that link.i; j / is not needed. The simple homogeneous netwok pesented in Section has that c ij D 2 ; 8i j, and i D ; 8i. In a netwok with identical nodes, it is natual to load-balance unifomly. But unifom load-balancing seems too estictive in a heteogeneous netwok because it does not take into account the diffeence in node capacities. A natual solution is to load-balance popotionally to the capacity of the intemediate node, i.e., ode i eceives faction i R of evey flow. This is a diect genealization fom unifom multicommodity flow in the homogeneous case to poduct multicommodity flow [16]. The equied link capacity is c ij D 2 i j =R [24]. We can futhe genealize VLB to allow any flow splitting atio and let some extenal objective to detemine the optimal atios. We intoduce a set of loadbalancing paametes p i such that p i 0, foalli, and P id1 p i D 1. An ingess node splits each flow accoding to fp i g and sends p i of evey flow to ode i. This gives us the feedom of, fo example, letting the lage nodes fowad moe taffic than the smalle nodes, o not using some of the nodes as intemediates (by setting the coesponding p i to zeo). If thee ae some objectives to be optimized, thee ae paametes (p i, i D 1;2;:::;) that can be tuned, but if moe feedom is needed, we can, fo example, let each flow have its own set of load-balancing paametes. We now find the equied link capacity. The fist-hop taffic on link.i; j / is the taffic initiated by ode i that is load-balanced to ode j,andtheateisatmost i p j. The second-hop taffic on the link is the taffic destined to ode j that is loadbalanced to ode i, and the ate is at most j p i. Theefoe the maximum amount of taffic on link.i; j / is i p j C j p i, which is also the equied capacity on link.i; j / c ij D i p j C j p i : (2.2) The equied (outgoing) inteconnection capacity of ode i is l i D X c ij D i C Rp i 2 i p i ; j Wj i and the total equied inteconnection capacity of the netwok is L D X l i D id1 X i;jwi j c ij D 2 R X i! i p i : (2.3)

6 24 R. Zhang-Shen Thus if we want to minimize the total equied inteconnection capacity of the netwok, we need to maximize P i i p i, subject to the constaints on p i. The optimal solution has the fom that p i >0only if i D max j j, i.e., taffic is only load-balanced to the lagest node(s). Thus we have min L D 2.R max i /: (2.4) i We can show that Equation (2.4) is also the minimum total inteconnection capacity equied by any netwok to suppot all taffic matices [24], and hence is the necessay and sufficient condition. One netwok that minimizes the total equied inteconnection capacity is a sta with ode 1 at the cente, i.e., p 1 D 1, p i D 0 fo i 2. In the case whee all nodes have the same capacity, to achieve minimum L, p i can take any non-negative values as long as they sum up to 1. Thus the unifom load-balancing pesented in Section is optimal in tems of total equied inteconnection capacity. Splitting flows popotional to node capacities, i.e., p i D i =R, is not optimal when nodes have diffeent capacities. In ode to use the minimum amount of inteconnection capacity, only nodes with the lagest capacity can act as intemediate nodes. This can be limiting, especially if only one node has the lagest capacity, because a sta netwok is not good fo fault toleance. A sta netwok is efficient but not balanced, because the cente node acts as the intemediate node fo all taffic. We need a scheme that is not only efficient but also balanced, and we found one by minimizing the netwok fanout. The fanout of node i is f i D l i i, the atio of node i s inteconnection capacity to its node capacity. Since the inteconnection capacity is used both fo sending taffic oiginated fom the node and fo fowading taffic fo othe nodes, the fanout measues the amount of esponsibility the node has to fowad othe nodes taffic elative to its size. If the fanouts of the two nodes ae the same, then the lage node fowads moe taffic, which is a desied popety. Thus, to have a balanced netwok, we minimize the maximum fanout ove all nodes, which esults in all nodes having equal fanout. The esulting load-balancing paametes and fanout ae p i D i R 2 P i k ; k R 2 k f i D 1 C i D 1;2;:::;; 1 P j ; 8i: j D1 R 2 j The optimal load-balancing paametes ae almost popotional to the node capacities: p i / i R 2 i. The paamete p i is a stictly inceasing function of i. Theefoe a lage node has geate esponsibility fo fowading taffic. We can futhe show that the total inteconnection capacity used in this scheme is no moe than 1 2.p 2 C 1/ D 1:207 times the minimum total capacity [24]. Thus, the scheme of minimizing maximum fanout is not only balanced, but also efficient. One can of couse choose to optimize othe citeia that ae moe suitable fo the paticula situation. Kodialam et al. [11] assume that the undelying physical links have fixed capacities, constaining the logical link capacities. They give efficient

7 2 Valiant Load-Balancing: Building etwoks That Can Suppot All Taffic Matices 25 algoithms fo calculating o appoximating the optimal load-balancing atios which allow the netwok to suppot the most taffic. Shephed and Winze [16] use a ealistic cost model that takes into account both the fibe cost and the equipment cost. They opt to load-balance to only a subset of the nodes so that a full mesh is not needed. 2.3 Fault-Toleance in a VLB etwok Most netwoks need to accommodate planned and unplanned inteuptions. As we show in this section, VLB netwoks, due to thei ich connectivity and path divesity, have many advantages in tems of toleating failues, such as All of the woking paths between a pai of nodes ae used all the time, and flows ae load-balanced acoss all woking paths. Most othe schemes equie potection paths that ae idle duing nomal opeation. All paths ae used all the time, so thee is no need to set up new paths upon failue. In ode to potect against k failues, 1 the faction of exta capacity equied is appoximately k. This is extemely efficient compaed to othe fault toleance schemes. VLB natually potects against multiple failues. One can decide duing design what failue scenaios the netwok should toleate, such as k abitay link o node failues o a paticula set of failue pattens. When thee ae no failues, each flow is load-balanced ove all the paths accoding to some atio. The same mechanism can be used when thee ae failues and some flows have fewe than woking paths. We assume that a node keeps tack of the available paths fo each flow oiginating fom it, and load-balances each flow ove the available paths, accoding to the same atios. The netwok is self-healing because the paths ae known pio to failues and taffic can be eouted as soon as failue is detected. We deive how much link capacity is needed so as to suppot all taffic matices unde k n abitay node failues 2 and k l abitay link failues. Fo simplicity, we assume the simple netwokmodelwheeall nodeshave the same capacity.dueto the symmety of the topology, all links in the fault-toleant netwok have the same capacity equiement, epesented by C.;k n ;k l /. We fist conside node failues and link failues sepaately and then combine them. By definition, C.; 0; 0/ D 2. ode failues ae elatively easy to analyze. When a node fails, it takes down all the links connecting to it and stops sending o eceiving taffic. The netwok becomes an. 1/-node full-mesh netwok, and a link capacity of is sufficient We focus on failues in the logical topology, and since seveal logical links can shae a physical link, a physical failue can coespond to multiple logical failues. 2 If a node fails, we discad the taffic oiginating fom o teminating at this node.

8 26 R. Zhang-Shen fo suppoting all taffic matices. In geneal, when thee ae k node failues, the netwok becomes an. k/-node full mesh, so the link capacity equied to toleate k node failues is C.;k;0/ D C. k; 0; 0/ D 2 k : (2.5) Link failues ae a little moe complicated, as they can destoy the symmety of the topology, so that it is no longe a full mesh. We only conside the wost-case failue scenaios (advesaial link failues) hee. We omit the details and only give the final esult. The amount of capacity equied on each link to toleate k abitay link failues, fo 1 k 2,is 8 < 2 C k D 1 C.;0;k/ D : k 1 C k D 2 o 2;o 6 1 othewise: 2 k (2.6) Fo k 1, the netwok becomes disconnected in the wost-case failue scenaio, theefoe we cannot guaantee a congestion-fee netwok. The significance of Equation (2.6) is that all enties on the ight-hand side ae vey close to 2 k. Figue 2.2 plots Equation (2.6)fo D 50, and as we can see, the cuve is vey flat fo small values of k. ow assume that thee ae k n node failues and k l link failues. This is equivalent to having k l link failues in a. k n /-node netwok. So C.;k n ;k l / 2 k n k l D 2 k ; (2.7) whee k D k n C k l. This means that the cuve fo Equation (2.7) is oughly the same as that fo Equation (2.6), shown in Figue 2.2. So we conclude that in a VLB netwok, a small amount of ove-povisioning goes a long way to make the netwok Fig. 2.2 The equied link capacity vs. the numbe of link failues in a 50-node netwok Requied capacity of each link (multiples of ) umbe of link failues, k

9 2 Valiant Load-Balancing: Building etwoks That Can Suppot All Taffic Matices 27 fault toleant. Fo example, if the links in a 50-node netwok ae ove-povisioned by just about 11%, the netwok can toleate any five (node o link) failues. 2.4 VLB fo Peeing Taffic Today, most congestion in Intenet backbone takes place on the peeing links connecting them. When a link between two netwoks is congested, vey likely some othe peeing links between these two netwoks ae lightly loaded. That is, the peeing links between two netwoks ae usually not evenly utilized. We popose to use a technique simila to VLB to oute peeing taffic, so as to eliminate congestion on the peeing links. We show that if taffic is load-balanced ove all the peeing links between two netwoks, thee will be no congestion as long as the total peeing capacity is geate than the total peeing taffic. Even though the two netwoks using VLB to oute thei peeing taffic do not need to use VLB intenally, we assume they do and analyze how the peeing scheme affects how the netwoks un VLB. Suppose two VLB netwoks ae connected by a subset of thei nodes (the peeing nodes), as shown in Figue 2.3. Fo the ease of desciption, we use the same numbeing fo the peeing nodes in both netwoks. (ote that this convention is diffeent fom Section 2.2.) The taffic exchanged between the two netwoks is called peeing taffic and assume the total amount is no moe than R p in each diection. In the netwok of nodes, we intoduce the peeing load-balancing paametes q i ;i D 1;2;:::;, such that a potion q i of the peeing taffic between the two netwoks is exchanged at node i. atually, q i D 0 if i is not a peeing node. The peeing load-balancing paametes, q i, togethe with the maximum peeing taffic between the two netwoks, R p, detemine the sizes of the peeing links: the equied capacity of the peeing link at node i is R p q i. Suppose the peeing links have the equied capacities, then if the peeing taffic is load-balanced acoss the peeing links accoding to the popotions q i, and the total amount of peeing taffic between the two netwoks does not exceed R p, thee will be no congestion on the peeing links. etwok 1 R p etwok 2 q q q 4 4 Fig. 2.3 Two VLB netwoks connect at a set of peeing nodes. The total amount of taffic exchanged between the two netwoks is no moe than R p, and a potion q i of the peeing taffic is exchanged at node i.

10 28 R. Zhang-Shen The exta equiement of outing peeing taffic may esult in highe capacity equiements inside the netwoks. If we teat the peeing taffic that oiginates fom the netwok as taffic destined to the peeing nodes, and the peeing taffic that entes the netwok as taffic oiginated fom the peeing nodes, then the peeing taffic may have to tavese two hops in each netwok. Altenatively, we can load-balance peeing taffic ove the peeing points only, instead of all the nodes. Thus, we equie that peeing taffic taveses at most one hop in each netwok, and at most two hops altogethe. Suppose the peeing load-balancing paametes ae fixed, fo example, though negotiation between the two netwoks. Then we can vay p i, i.e., how non-peeing taffic is outed intenally, to minimize the equied link capacities. We obseve that R p is likely to be bigge than the node capacities i. R p is the total amount of taffic the two netwok exchanges and can be a lage potion of the netwok s total taffic R, while the node capacities ae likely to make up only a small faction of R, onthe ode of R. If we assume that R p i fo all i,thenwehave c ij D i max.p j ;q j / C j max.p i ;q i /; and the minimum c ij is achieved fo all links when p i D q i fo all i. So the optimal capacity allocation in a netwok with peeing taffic is c ij D i q j C j q i : (2.8) Since q i is zeo if node i is a non-peeing node, c ij D 0 if both ode i and ode j ae non-peeing nodes. The netwok is now a two-tieed one: in the cente is a full mesh connecting the peeing nodes, and on the edge ae the non-peeing nodes, each connected to all the cente nodes. Setting local load-balancing paametes to be the same as peeing load-balancing paametes means that only the peeing nodes will seve as intemediate nodes to fowad taffic. Peeing nodes ae often the lagest nodes in the netwok, so they should have lage esponsibilities in fowading taffic. The analysis shows that the optimal way is to only let the peeing nodes fowad taffic. This has the additional benefits of equiing fewe links and educing netwok complexity. 2.5 Discussions Using VLB to oute taffic in a netwok has many advantages, such as efficiency and fast failue ecovey. Thee ae some open questions as well. Some examples ae Sending packets though two hops may incease the popagation delay, while sending packets though the diect link may cause inceased capacity equiement. Thee should be a tadeoff between packet delay and equied capacity.

11 2 Valiant Load-Balancing: Building etwoks That Can Suppot All Taffic Matices 29 Diffeent paths that a flow taveses may have diffeent delays, and hence packets in the flow may not aive at the destination in the oiginal ode. This is usually not a poblem in the Intenet, since a flow may consist of many application-level flows, and can be split accodingly fo load-balancing. But it can be a poblem if a flow cannot be easily split. It is unclea whethe VLB can be incementally deployed in a netwok. Despite the challenges, VLB is apidly gaining populaity in netwoking eseach. This suvey pape can be a stating point fo those inteested in applying VLB to othe situations. Refeences 1. R. Aleliunas. Randomized paallel communication (peliminay vesion). In PODC 82: Poceedings of the fist ACM SIGACT-SIGOPS symposium on Pinciples of distibuted computing, pages 60 72, K. Agyaki, S. Baset, B.-G. Chun, K. Fall, G. Iannaccone, A. Knies, E. Kohle, M. Manesh, S. edevschi, and S. Ratnasamy. Can softwae outes scale? In PRESTO 08: Poceedings of the ACM wokshop on Pogammable outes fo extensible sevices of tomoow, pages 21 26, P. Benasconi, J. Gipp, D. eilson, J. Simsaian, D. Stiliadis, A. Vama, and M. Zingibl. Achitectue of an integated oute inteconnected spectally (IRIS). High Pefomance Switching and Routing, 2006 Wokshop on, pages 8 pp., June C.-S. Chang, D.-S. Lee, and Y.-S. Jou. Load balanced Bikhoff-von eumann switches, Pat I: One-stage buffeing. Compute Communications, 25(6): , C.-S. Chang, D.-S. Lee, and C.-M. Lien. Load balanced Bikhoff-von eumann switches, Pat II: Multi-stage buffeing. Compute Communications, 25(6): , A. Geenbeg, P. Lahii, D. A. Maltz, P. Patel, and S. Sengupta. Towads a next geneation data cente achitectue: scalability and commoditization. In PRESTO 08: Poceedings of the ACM wokshop on Pogammable outes fo extensible sevices of tomoow, pages 57 62, M. Henion, K. Schodi, D. Boettle, M. De Some, and M. Dieudonne. Switching netwok achitectue fo ATM based boadband communications. Switching Symposium, XIII Intenational, 5:1 8, I. Keslassy, C.-S. Chang,. McKeown, and D.-S. Lee. Optimal load-balancing. In Poc. IEEE IFOCOM, I. Keslassy, S.-T. Chuang, K. Yu, D. Mille, M. Hoowitz, O. Solgaad, and. McKeown. Scaling Intenet outes using optics. Poceedings of ACM SIGCOMM 03, Compute Communication Review, 33(4): , Octobe I. Keslassy, M. Kodialam, T. Lakshman, and D. Stiliadis. Scheduling schemes fo delay gaphs with applications to optical packet netwoks. High Pefomance Switching and Routing (HPSR), pages , M. Kodialam, T. V. Lakshman, J. B. Olin, and S. Sengupta. A Vesatile Scheme fo Routing Highly Vaiable Taffic in Sevice Ovelays and IP Backbones. In Poc. IEEE IFOCOM, Apil M. Kodialam, T. V. Lakshman, and S. Sengupta. Efficient and obust outing of highly vaiable taffic. In Hotets III, ovembe D. Mita and R. A. Cieslak. Randomized paallel communications on an extension of the omega netwok. J. ACM, 34(4): , H. agesh, V. Poosala, V. Kuma, P. Winze, and M. Zingibl. Load-balanced achitectue fo dynamic taffic. In Optical Fibe Communication Confeence, Mach 2005.

12 30 R. Zhang-Shen 15. R. Pasad, P. Winze, S. Bost, and M. Thottan. Queuing delays in andomized load balanced netwoks. In Poc. IEEE IFOCOM, May F. B. Shephed and P. J. Winze. Selective andomized load balancing and mesh netwoks with changing demands. Jounal of Optical etwoking, 5: , A. Singh. Load-Balanced Routing in Inteconnection etwoks. PhD thesis, Depatment of Electical Engineeing, Stanfod Univesity, A. Singh, W. J. Dally, B. Towles, and A. K. Gupta. Locality-peseving andomized oblivious outing on tous netwoks. In SPAA 02: Poceedings of the fouteenth annual ACM symposium on paallel algoithms and achitectues, pages 9 13, L. G. Valiant. A scheme fo fast paallel communication. SIAM Jounal on Computing, 11(2): , R. van Haalen, R. Malhota, and A. de Hee. Optimized outing fo poviding Ethenet LA sevices. Communications Magazine, IEEE, 43(11): , ov P. J. Winze, F. B. Shephed, P. Oswald, and M. Zingibl. Robust netwok design and selective andomized load balancing. 31st Euopean Confeence on Optical Communication (ECOC), 1:23 24, Septembe R. Zhang-Shen, M. Kodialam, and T. V. Lakshman. Achieving bounded blocking in cicuitswitched netwoks. IEEE IFOCOM 2006, pages 1 9, Apil R. Zhang-Shen and. McKeown. Designing a Pedictable Intenet Backbone etwok. In Hotets III, ovembe R. Zhang-Shen and. McKeown. Designing a pedictable Intenet backbone with Valiant Load-Balancing. Thiteenth Intenational Wokshop on Quality of Sevice (IWQoS), R. Zhang-Shen and. McKeown. Designing a Fault-Toleant etwok Using Valiant Load- Balancing. Poc. IEEE IFOCOM, pages , Apil R. Zhang-Shen and. McKeown. Guaanteeing Quality of Sevice to Peeing Taffic. Poc. IEEE IFOCOM, pages , Apil 2008.

13

Effect of Contention Window on the Performance of IEEE 802.11 WLANs

Effect of Contention Window on the Performance of IEEE 802.11 WLANs Effect of Contention Window on the Pefomance of IEEE 82.11 WLANs Yunli Chen and Dhama P. Agawal Cente fo Distibuted and Mobile Computing, Depatment of ECECS Univesity of Cincinnati, OH 45221-3 {ychen,

More information

Questions & Answers Chapter 10 Software Reliability Prediction, Allocation and Demonstration Testing

Questions & Answers Chapter 10 Software Reliability Prediction, Allocation and Demonstration Testing M13914 Questions & Answes Chapte 10 Softwae Reliability Pediction, Allocation and Demonstation Testing 1. Homewok: How to deive the fomula of failue ate estimate. λ = χ α,+ t When the failue times follow

More information

Software Engineering and Development

Software Engineering and Development I T H E A 67 Softwae Engineeing and Development SOFTWARE DEVELOPMENT PROCESS DYNAMICS MODELING AS STATE MACHINE Leonid Lyubchyk, Vasyl Soloshchuk Abstact: Softwae development pocess modeling is gaining

More information

How To Use A Network On A Network With A Powerline (Lan) On A Pcode (Lan On Alan) (Lan For Acedo) (Moe) (Omo) On An Ipo) Or Ipo (

How To Use A Network On A Network With A Powerline (Lan) On A Pcode (Lan On Alan) (Lan For Acedo) (Moe) (Omo) On An Ipo) Or Ipo ( Hubs, Bidges, and Switches Used fo extending LANs in tems of geogaphical coveage, numbe of nodes, administation capabilities, etc. Diffe in egads to: m collision domain isolation m laye at which they opeate

More information

The transport performance evaluation system building of logistics enterprises

The transport performance evaluation system building of logistics enterprises Jounal of Industial Engineeing and Management JIEM, 213 6(4): 194-114 Online ISSN: 213-953 Pint ISSN: 213-8423 http://dx.doi.og/1.3926/jiem.784 The tanspot pefomance evaluation system building of logistics

More information

ON THE (Q, R) POLICY IN PRODUCTION-INVENTORY SYSTEMS

ON THE (Q, R) POLICY IN PRODUCTION-INVENTORY SYSTEMS ON THE R POLICY IN PRODUCTION-INVENTORY SYSTEMS Saifallah Benjaafa and Joon-Seok Kim Depatment of Mechanical Engineeing Univesity of Minnesota Minneapolis MN 55455 Abstact We conside a poduction-inventoy

More information

HEALTHCARE INTEGRATION BASED ON CLOUD COMPUTING

HEALTHCARE INTEGRATION BASED ON CLOUD COMPUTING U.P.B. Sci. Bull., Seies C, Vol. 77, Iss. 2, 2015 ISSN 2286-3540 HEALTHCARE INTEGRATION BASED ON CLOUD COMPUTING Roxana MARCU 1, Dan POPESCU 2, Iulian DANILĂ 3 A high numbe of infomation systems ae available

More information

AN IMPLEMENTATION OF BINARY AND FLOATING POINT CHROMOSOME REPRESENTATION IN GENETIC ALGORITHM

AN IMPLEMENTATION OF BINARY AND FLOATING POINT CHROMOSOME REPRESENTATION IN GENETIC ALGORITHM AN IMPLEMENTATION OF BINARY AND FLOATING POINT CHROMOSOME REPRESENTATION IN GENETIC ALGORITHM Main Golub Faculty of Electical Engineeing and Computing, Univesity of Zageb Depatment of Electonics, Micoelectonics,

More information

STUDENT RESPONSE TO ANNUITY FORMULA DERIVATION

STUDENT RESPONSE TO ANNUITY FORMULA DERIVATION Page 1 STUDENT RESPONSE TO ANNUITY FORMULA DERIVATION C. Alan Blaylock, Hendeson State Univesity ABSTRACT This pape pesents an intuitive appoach to deiving annuity fomulas fo classoom use and attempts

More information

Cloud Service Reliability: Modeling and Analysis

Cloud Service Reliability: Modeling and Analysis Cloud Sevice eliability: Modeling and Analysis Yuan-Shun Dai * a c, Bo Yang b, Jack Dongaa a, Gewei Zhang c a Innovative Computing Laboatoy, Depatment of Electical Engineeing & Compute Science, Univesity

More information

YARN PROPERTIES MEASUREMENT: AN OPTICAL APPROACH

YARN PROPERTIES MEASUREMENT: AN OPTICAL APPROACH nd INTERNATIONAL TEXTILE, CLOTHING & ESIGN CONFERENCE Magic Wold of Textiles Octobe 03 d to 06 th 004, UBROVNIK, CROATIA YARN PROPERTIES MEASUREMENT: AN OPTICAL APPROACH Jana VOBOROVA; Ashish GARG; Bohuslav

More information

Comparing Availability of Various Rack Power Redundancy Configurations

Comparing Availability of Various Rack Power Redundancy Configurations Compaing Availability of Vaious Rack Powe Redundancy Configuations White Pape 48 Revision by Victo Avela > Executive summay Tansfe switches and dual-path powe distibution to IT equipment ae used to enhance

More information

An Infrastructure Cost Evaluation of Single- and Multi-Access Networks with Heterogeneous Traffic Density

An Infrastructure Cost Evaluation of Single- and Multi-Access Networks with Heterogeneous Traffic Density An Infastuctue Cost Evaluation of Single- and Multi-Access Netwoks with Heteogeneous Taffic Density Andes Fuuskä and Magnus Almgen Wieless Access Netwoks Eicsson Reseach Kista, Sweden [andes.fuuska, magnus.almgen]@eicsson.com

More information

Chapter 3 Savings, Present Value and Ricardian Equivalence

Chapter 3 Savings, Present Value and Ricardian Equivalence Chapte 3 Savings, Pesent Value and Ricadian Equivalence Chapte Oveview In the pevious chapte we studied the decision of households to supply hous to the labo maket. This decision was a static decision,

More information

An Approach to Optimized Resource Allocation for Cloud Simulation Platform

An Approach to Optimized Resource Allocation for Cloud Simulation Platform An Appoach to Optimized Resouce Allocation fo Cloud Simulation Platfom Haitao Yuan 1, Jing Bi 2, Bo Hu Li 1,3, Xudong Chai 3 1 School of Automation Science and Electical Engineeing, Beihang Univesity,

More information

Efficient Redundancy Techniques for Latency Reduction in Cloud Systems

Efficient Redundancy Techniques for Latency Reduction in Cloud Systems Efficient Redundancy Techniques fo Latency Reduction in Cloud Systems 1 Gaui Joshi, Emina Soljanin, and Gegoy Wonell Abstact In cloud computing systems, assigning a task to multiple seves and waiting fo

More information

A Comparative Analysis of Data Center Network Architectures

A Comparative Analysis of Data Center Network Architectures A Compaative Analysis of Data Cente Netwok Achitectues Fan Yao, Jingxin Wu, Guu Venkataamani, Suesh Subamaniam Depatment of Electical and Compute Engineeing, The Geoge Washington Univesity, Washington,

More information

High Availability Replication Strategy for Deduplication Storage System

High Availability Replication Strategy for Deduplication Storage System Zhengda Zhou, Jingli Zhou College of Compute Science and Technology, Huazhong Univesity of Science and Technology, *, zhouzd@smail.hust.edu.cn jlzhou@mail.hust.edu.cn Abstact As the amount of digital data

More information

Optimizing Content Retrieval Delay for LT-based Distributed Cloud Storage Systems

Optimizing Content Retrieval Delay for LT-based Distributed Cloud Storage Systems Optimizing Content Retieval Delay fo LT-based Distibuted Cloud Stoage Systems Haifeng Lu, Chuan Heng Foh, Yonggang Wen, and Jianfei Cai School of Compute Engineeing, Nanyang Technological Univesity, Singapoe

More information

An Epidemic Model of Mobile Phone Virus

An Epidemic Model of Mobile Phone Virus An Epidemic Model of Mobile Phone Vius Hui Zheng, Dong Li, Zhuo Gao 3 Netwok Reseach Cente, Tsinghua Univesity, P. R. China zh@tsinghua.edu.cn School of Compute Science and Technology, Huazhong Univesity

More information

An Introduction to Omega

An Introduction to Omega An Intoduction to Omega Con Keating and William F. Shadwick These distibutions have the same mean and vaiance. Ae you indiffeent to thei isk-ewad chaacteistics? The Finance Development Cente 2002 1 Fom

More information

Adaptive Queue Management with Restraint on Non-Responsive Flows

Adaptive Queue Management with Restraint on Non-Responsive Flows Adaptive Queue Management wi Restaint on Non-Responsive Flows Lan Li and Gyungho Lee Depatment of Electical and Compute Engineeing Univesity of Illinois at Chicago 85 S. Mogan Steet Chicago, IL 667 {lli,

More information

Comparing Availability of Various Rack Power Redundancy Configurations

Comparing Availability of Various Rack Power Redundancy Configurations Compaing Availability of Vaious Rack Powe Redundancy Configuations By Victo Avela White Pape #48 Executive Summay Tansfe switches and dual-path powe distibution to IT equipment ae used to enhance the availability

More information

Concept and Experiences on using a Wiki-based System for Software-related Seminar Papers

Concept and Experiences on using a Wiki-based System for Software-related Seminar Papers Concept and Expeiences on using a Wiki-based System fo Softwae-elated Semina Papes Dominik Fanke and Stefan Kowalewski RWTH Aachen Univesity, 52074 Aachen, Gemany, {fanke, kowalewski}@embedded.wth-aachen.de,

More information

Ilona V. Tregub, ScD., Professor

Ilona V. Tregub, ScD., Professor Investment Potfolio Fomation fo the Pension Fund of Russia Ilona V. egub, ScD., Pofesso Mathematical Modeling of Economic Pocesses Depatment he Financial Univesity unde the Govenment of the Russian Fedeation

More information

The Role of Gravity in Orbital Motion

The Role of Gravity in Orbital Motion ! The Role of Gavity in Obital Motion Pat of: Inquiy Science with Datmouth Developed by: Chistophe Caoll, Depatment of Physics & Astonomy, Datmouth College Adapted fom: How Gavity Affects Obits (Ohio State

More information

An Efficient Group Key Agreement Protocol for Ad hoc Networks

An Efficient Group Key Agreement Protocol for Ad hoc Networks An Efficient Goup Key Ageement Potocol fo Ad hoc Netwoks Daniel Augot, Raghav haska, Valéie Issany and Daniele Sacchetti INRIA Rocquencout 78153 Le Chesnay Fance {Daniel.Augot, Raghav.haska, Valéie.Issany,

More information

Scheduling Hadoop Jobs to Meet Deadlines

Scheduling Hadoop Jobs to Meet Deadlines Scheduling Hadoop Jobs to Meet Deadlines Kamal Kc, Kemafo Anyanwu Depatment of Compute Science Noth Caolina State Univesity {kkc,kogan}@ncsu.edu Abstact Use constaints such as deadlines ae impotant equiements

More information

Modeling and Verifying a Price Model for Congestion Control in Computer Networks Using PROMELA/SPIN

Modeling and Verifying a Price Model for Congestion Control in Computer Networks Using PROMELA/SPIN Modeling and Veifying a Pice Model fo Congestion Contol in Compute Netwoks Using PROMELA/SPIN Clement Yuen and Wei Tjioe Depatment of Compute Science Univesity of Toonto 1 King s College Road, Toonto,

More information

FXA 2008. Candidates should be able to : Describe how a mass creates a gravitational field in the space around it.

FXA 2008. Candidates should be able to : Describe how a mass creates a gravitational field in the space around it. Candidates should be able to : Descibe how a mass ceates a gavitational field in the space aound it. Define gavitational field stength as foce pe unit mass. Define and use the peiod of an object descibing

More information

Optimal Peer Selection in a Free-Market Peer-Resource Economy

Optimal Peer Selection in a Free-Market Peer-Resource Economy Optimal Pee Selection in a Fee-Maket Pee-Resouce Economy Micah Adle, Rakesh Kuma, Keith Ross, Dan Rubenstein, David Tune and David D Yao Dept of Compute Science Univesity of Massachusetts Amhest, MA; Email:

More information

INITIAL MARGIN CALCULATION ON DERIVATIVE MARKETS OPTION VALUATION FORMULAS

INITIAL MARGIN CALCULATION ON DERIVATIVE MARKETS OPTION VALUATION FORMULAS INITIAL MARGIN CALCULATION ON DERIVATIVE MARKETS OPTION VALUATION FORMULAS Vesion:.0 Date: June 0 Disclaime This document is solely intended as infomation fo cleaing membes and othes who ae inteested in

More information

Research on Risk Assessment of the Transformer Based on Life Cycle Cost

Research on Risk Assessment of the Transformer Based on Life Cycle Cost ntenational Jounal of Smat Gid and lean Enegy eseach on isk Assessment of the Tansfome Based on Life ycle ost Hui Zhou a, Guowei Wu a, Weiwei Pan a, Yunhe Hou b, hong Wang b * a Zhejiang Electic Powe opoation,

More information

Things to Remember. r Complete all of the sections on the Retirement Benefit Options form that apply to your request.

Things to Remember. r Complete all of the sections on the Retirement Benefit Options form that apply to your request. Retiement Benefit 1 Things to Remembe Complete all of the sections on the Retiement Benefit fom that apply to you equest. If this is an initial equest, and not a change in a cuent distibution, emembe to

More information

Continuous Compounding and Annualization

Continuous Compounding and Annualization Continuous Compounding and Annualization Philip A. Viton Januay 11, 2006 Contents 1 Intoduction 1 2 Continuous Compounding 2 3 Pesent Value with Continuous Compounding 4 4 Annualization 5 5 A Special Poblem

More information

Data Center Demand Response: Avoiding the Coincident Peak via Workload Shifting and Local Generation

Data Center Demand Response: Avoiding the Coincident Peak via Workload Shifting and Local Generation (213) 1 28 Data Cente Demand Response: Avoiding the Coincident Peak via Wokload Shifting and Local Geneation Zhenhua Liu 1, Adam Wieman 1, Yuan Chen 2, Benjamin Razon 1, Niangjun Chen 1 1 Califonia Institute

More information

est using the formula I = Prt, where I is the interest earned, P is the principal, r is the interest rate, and t is the time in years.

est using the formula I = Prt, where I is the interest earned, P is the principal, r is the interest rate, and t is the time in years. 9.2 Inteest Objectives 1. Undestand the simple inteest fomula. 2. Use the compound inteest fomula to find futue value. 3. Solve the compound inteest fomula fo diffeent unknowns, such as the pesent value,

More information

Automatic Testing of Neighbor Discovery Protocol Based on FSM and TTCN*

Automatic Testing of Neighbor Discovery Protocol Based on FSM and TTCN* Automatic Testing of Neighbo Discovey Potocol Based on FSM and TTCN* Zhiliang Wang, Xia Yin, Haibin Wang, and Jianping Wu Depatment of Compute Science, Tsinghua Univesity Beijing, P. R. China, 100084 Email:

More information

The impact of migration on the provision. of UK public services (SRG.10.039.4) Final Report. December 2011

The impact of migration on the provision. of UK public services (SRG.10.039.4) Final Report. December 2011 The impact of migation on the povision of UK public sevices (SRG.10.039.4) Final Repot Decembe 2011 The obustness The obustness of the analysis of the is analysis the esponsibility is the esponsibility

More information

Reduced Pattern Training Based on Task Decomposition Using Pattern Distributor

Reduced Pattern Training Based on Task Decomposition Using Pattern Distributor > PNN05-P762 < Reduced Patten Taining Based on Task Decomposition Using Patten Distibuto Sheng-Uei Guan, Chunyu Bao, and TseNgee Neo Abstact Task Decomposition with Patten Distibuto (PD) is a new task

More information

Give me all I pay for Execution Guarantees in Electronic Commerce Payment Processes

Give me all I pay for Execution Guarantees in Electronic Commerce Payment Processes Give me all I pay fo Execution Guaantees in Electonic Commece Payment Pocesses Heiko Schuldt Andei Popovici Hans-Jög Schek Email: Database Reseach Goup Institute of Infomation Systems ETH Zentum, 8092

More information

Load Balancing in Processor Sharing Systems

Load Balancing in Processor Sharing Systems Load Balancing in ocesso Shaing Systems Eitan Altman INRIA Sophia Antipolis 2004, oute des Lucioles 06902 Sophia Antipolis, Fance altman@sophia.inia.f Utzi Ayesta LAAS-CNRS Univesité de Toulouse 7, Avenue

More information

Load Balancing in Processor Sharing Systems

Load Balancing in Processor Sharing Systems Load Balancing in ocesso Shaing Systems Eitan Altman INRIA Sophia Antipolis 2004, oute des Lucioles 06902 Sophia Antipolis, Fance altman@sophia.inia.f Utzi Ayesta LAAS-CNRS Univesité de Toulouse 7, Avenue

More information

Timing Synchronization in High Mobility OFDM Systems

Timing Synchronization in High Mobility OFDM Systems Timing Synchonization in High Mobility OFDM Systems Yasamin Mostofi Depatment of Electical Engineeing Stanfod Univesity Stanfod, CA 94305, USA Email: yasi@wieless.stanfod.edu Donald C. Cox Depatment of

More information

Strength Analysis and Optimization Design about the key parts of the Robot

Strength Analysis and Optimization Design about the key parts of the Robot Intenational Jounal of Reseach in Engineeing and Science (IJRES) ISSN (Online): 2320-9364, ISSN (Pint): 2320-9356 www.ijes.og Volume 3 Issue 3 ǁ Mach 2015 ǁ PP.25-29 Stength Analysis and Optimization Design

More information

IBM Research Smarter Transportation Analytics

IBM Research Smarter Transportation Analytics IBM Reseach Smate Tanspotation Analytics Laua Wynte PhD, Senio Reseach Scientist, IBM Watson Reseach Cente lwynte@us.ibm.com INSTRUMENTED We now have the ability to measue, sense and see the exact condition

More information

Episode 401: Newton s law of universal gravitation

Episode 401: Newton s law of universal gravitation Episode 401: Newton s law of univesal gavitation This episode intoduces Newton s law of univesal gavitation fo point masses, and fo spheical masses, and gets students pactising calculations of the foce

More information

Financing Terms in the EOQ Model

Financing Terms in the EOQ Model Financing Tems in the EOQ Model Habone W. Stuat, J. Columbia Business School New Yok, NY 1007 hws7@columbia.edu August 6, 004 1 Intoduction This note discusses two tems that ae often omitted fom the standad

More information

Questions for Review. By buying bonds This period you save s, next period you get s(1+r)

Questions for Review. By buying bonds This period you save s, next period you get s(1+r) MACROECONOMICS 2006 Week 5 Semina Questions Questions fo Review 1. How do consumes save in the two-peiod model? By buying bonds This peiod you save s, next peiod you get s() 2. What is the slope of a consume

More information

THE DISTRIBUTED LOCATION RESOLUTION PROBLEM AND ITS EFFICIENT SOLUTION

THE DISTRIBUTED LOCATION RESOLUTION PROBLEM AND ITS EFFICIENT SOLUTION IADIS Intenational Confeence Applied Computing 2006 THE DISTRIBUTED LOCATION RESOLUTION PROBLEM AND ITS EFFICIENT SOLUTION Jög Roth Univesity of Hagen 58084 Hagen, Gemany Joeg.Roth@Fenuni-hagen.de ABSTRACT

More information

Chris J. Skinner The probability of identification: applying ideas from forensic statistics to disclosure risk assessment

Chris J. Skinner The probability of identification: applying ideas from forensic statistics to disclosure risk assessment Chis J. Skinne The pobability of identification: applying ideas fom foensic statistics to disclosue isk assessment Aticle (Accepted vesion) (Refeeed) Oiginal citation: Skinne, Chis J. (2007) The pobability

More information

An Analysis of Manufacturer Benefits under Vendor Managed Systems

An Analysis of Manufacturer Benefits under Vendor Managed Systems An Analysis of Manufactue Benefits unde Vendo Managed Systems Seçil Savaşaneil Depatment of Industial Engineeing, Middle East Technical Univesity, 06531, Ankaa, TURKEY secil@ie.metu.edu.t Nesim Ekip 1

More information

Tracking/Fusion and Deghosting with Doppler Frequency from Two Passive Acoustic Sensors

Tracking/Fusion and Deghosting with Doppler Frequency from Two Passive Acoustic Sensors Tacking/Fusion and Deghosting with Dopple Fequency fom Two Passive Acoustic Sensos Rong Yang, Gee Wah Ng DSO National Laboatoies 2 Science Pak Dive Singapoe 11823 Emails: yong@dso.og.sg, ngeewah@dso.og.sg

More information

Evaluating the impact of Blade Server and Virtualization Software Technologies on the RIT Datacenter

Evaluating the impact of Blade Server and Virtualization Software Technologies on the RIT Datacenter Evaluating the impact of and Vitualization Softwae Technologies on the RIT Datacente Chistophe M Butle Vitual Infastuctue Administato Rocheste Institute of Technology s Datacente Contact: chis.butle@it.edu

More information

An application of stochastic programming in solving capacity allocation and migration planning problem under uncertainty

An application of stochastic programming in solving capacity allocation and migration planning problem under uncertainty An application of stochastic pogamming in solving capacity allocation and migation planning poblem unde uncetainty Yin-Yann Chen * and Hsiao-Yao Fan Depatment of Industial Management, National Fomosa Univesity,

More information

UNIVERSIDAD DE CANTABRIA TESIS DOCTORAL

UNIVERSIDAD DE CANTABRIA TESIS DOCTORAL UNIVERSIDAD DE CANABRIA Depatamento de Ingenieía de Comunicaciones ESIS DOCORAL Cyogenic echnology in the Micowave Engineeing: Application to MIC and MMIC Vey Low Noise Amplifie Design Juan Luis Cano de

More information

Towards Realizing a Low Cost and Highly Available Datacenter Power Infrastructure

Towards Realizing a Low Cost and Highly Available Datacenter Power Infrastructure Towads Realizing a Low Cost and Highly Available Datacente Powe Infastuctue Siam Govindan, Di Wang, Lydia Chen, Anand Sivasubamaniam, and Bhuvan Ugaonka The Pennsylvania State Univesity. IBM Reseach Zuich

More information

Approximation Algorithms for Data Management in Networks

Approximation Algorithms for Data Management in Networks Appoximation Algoithms fo Data Management in Netwoks Chistof Kick Heinz Nixdof Institute and Depatment of Mathematics & Compute Science adebon Univesity Gemany kueke@upb.de Haald Räcke Heinz Nixdof Institute

More information

A Capacitated Commodity Trading Model with Market Power

A Capacitated Commodity Trading Model with Market Power A Capacitated Commodity Tading Model with Maket Powe Victo Matínez-de-Albéniz Josep Maia Vendell Simón IESE Business School, Univesity of Navaa, Av. Peason 1, 08034 Bacelona, Spain VAlbeniz@iese.edu JMVendell@iese.edu

More information

INVESTIGATION OF FLOW INSIDE AN AXIAL-FLOW PUMP OF GV IMP TYPE

INVESTIGATION OF FLOW INSIDE AN AXIAL-FLOW PUMP OF GV IMP TYPE 1 INVESTIGATION OF FLOW INSIDE AN AXIAL-FLOW PUMP OF GV IMP TYPE ANATOLIY A. YEVTUSHENKO 1, ALEXEY N. KOCHEVSKY 1, NATALYA A. FEDOTOVA 1, ALEXANDER Y. SCHELYAEV 2, VLADIMIR N. KONSHIN 2 1 Depatment of

More information

Financial Derivatives for Computer Network Capacity Markets with Quality-of-Service Guarantees

Financial Derivatives for Computer Network Capacity Markets with Quality-of-Service Guarantees Financial Deivatives fo Compute Netwok Capacity Makets with Quality-of-Sevice Guaantees Pette Pettesson pp@kth.se Febuay 2003 SICS Technical Repot T2003:03 Keywods Netwoking and Intenet Achitectue. Abstact

More information

SUPPORT VECTOR MACHINE FOR BANDWIDTH ANALYSIS OF SLOTTED MICROSTRIP ANTENNA

SUPPORT VECTOR MACHINE FOR BANDWIDTH ANALYSIS OF SLOTTED MICROSTRIP ANTENNA Intenational Jounal of Compute Science, Systems Engineeing and Infomation Technology, 4(), 20, pp. 67-7 SUPPORT VECTOR MACHIE FOR BADWIDTH AALYSIS OF SLOTTED MICROSTRIP ATEA Venmathi A.R. & Vanitha L.

More information

2 r2 θ = r2 t. (3.59) The equal area law is the statement that the term in parentheses,

2 r2 θ = r2 t. (3.59) The equal area law is the statement that the term in parentheses, 3.4. KEPLER S LAWS 145 3.4 Keple s laws You ae familia with the idea that one can solve some mechanics poblems using only consevation of enegy and (linea) momentum. Thus, some of what we see as objects

More information

Supplementary Material for EpiDiff

Supplementary Material for EpiDiff Supplementay Mateial fo EpiDiff Supplementay Text S1. Pocessing of aw chomatin modification data In ode to obtain the chomatin modification levels in each of the egions submitted by the use QDCMR module

More information

A framework for the selection of enterprise resource planning (ERP) system based on fuzzy decision making methods

A framework for the selection of enterprise resource planning (ERP) system based on fuzzy decision making methods A famewok fo the selection of entepise esouce planning (ERP) system based on fuzzy decision making methods Omid Golshan Tafti M.s student in Industial Management, Univesity of Yazd Omidgolshan87@yahoo.com

More information

Firstmark Credit Union Commercial Loan Department

Firstmark Credit Union Commercial Loan Department Fistmak Cedit Union Commecial Loan Depatment Thank you fo consideing Fistmak Cedit Union as a tusted souce to meet the needs of you business. Fistmak Cedit Union offes a wide aay of business loans and

More information

Peer-to-Peer File Sharing Game using Correlated Equilibrium

Peer-to-Peer File Sharing Game using Correlated Equilibrium Pee-to-Pee File Shaing Game using Coelated Equilibium Beibei Wang, Zhu Han, and K. J. Ray Liu Depatment of Electical and Compute Engineeing and Institute fo Systems Reseach, Univesity of Mayland, College

More information

STABILITY ANALYSIS IN MILLING BASED ON OPERATIONAL MODAL DATA 1. INTRODUCTION

STABILITY ANALYSIS IN MILLING BASED ON OPERATIONAL MODAL DATA 1. INTRODUCTION Jounal of Machine Engineeing, Vol. 11, No. 4, 211 Batosz POWALKA 1 Macin CHODZKO 1 Kzysztof JEMIELNIAK 2 milling, chatte, opeational modal analysis STABILITY ANALYSIS IN MILLING BASED ON OPERATIONAL MODAL

More information

Channel selection in e-commerce age: A strategic analysis of co-op advertising models

Channel selection in e-commerce age: A strategic analysis of co-op advertising models Jounal of Industial Engineeing and Management JIEM, 013 6(1):89-103 Online ISSN: 013-0953 Pint ISSN: 013-843 http://dx.doi.og/10.396/jiem.664 Channel selection in e-commece age: A stategic analysis of

More information

NBER WORKING PAPER SERIES FISCAL ZONING AND SALES TAXES: DO HIGHER SALES TAXES LEAD TO MORE RETAILING AND LESS MANUFACTURING?

NBER WORKING PAPER SERIES FISCAL ZONING AND SALES TAXES: DO HIGHER SALES TAXES LEAD TO MORE RETAILING AND LESS MANUFACTURING? NBER WORKING PAPER SERIES FISCAL ZONING AND SALES TAXES: DO HIGHER SALES TAXES LEAD TO MORE RETAILING AND LESS MANUFACTURING? Daia Bunes David Neumak Michelle J. White Woking Pape 16932 http://www.nbe.og/papes/w16932

More information

Intertemporal Macroeconomics

Intertemporal Macroeconomics Intetempoal Macoeconomics Genot Doppelhofe* May 2009 Fothcoming in J. McCombie and N. Allington (eds.), Cambidge Essays in Applied Economics, Cambidge UP This chapte eviews models of intetempoal choice

More information

Distributed Computing and Big Data: Hadoop and MapReduce

Distributed Computing and Big Data: Hadoop and MapReduce Distibuted Computing and Big Data: Hadoop and Map Bill Keenan, Diecto Tey Heinze, Achitect Thomson Reutes Reseach & Development Agenda R&D Oveview Hadoop and Map Oveview Use Case: Clusteing Legal Documents

More information

Multicriteria analysis in telecommunications

Multicriteria analysis in telecommunications Poceedings of the 37th Hawaii Intenational Confeence on System Sciences - 2004 Multiciteia analysis in telecommunications Janusz Ganat and Andze P. Wiezbicki National Institute of Telecommunications Szachowa

More information

METHODOLOGICAL APPROACH TO STRATEGIC PERFORMANCE OPTIMIZATION

METHODOLOGICAL APPROACH TO STRATEGIC PERFORMANCE OPTIMIZATION ETHODOOGICA APPOACH TO STATEGIC PEFOANCE OPTIIZATION ao Hell * Stjepan Vidačić ** Željo Gaača *** eceived: 4. 07. 2009 Peliminay communication Accepted: 5. 0. 2009 UDC 65.02.4 This pape pesents a matix

More information

Energy Efficient Cache Invalidation in a Mobile Environment

Energy Efficient Cache Invalidation in a Mobile Environment Enegy Efficient Cache Invalidation in a Mobile Envionment Naottam Chand, Ramesh Chanda Joshi, Manoj Misa Electonics & Compute Engineeing Depatment Indian Institute of Technology, Rookee - 247 667. INDIA

More information

Mining Relatedness Graphs for Data Integration

Mining Relatedness Graphs for Data Integration Mining Relatedness Gaphs fo Data Integation Jeemy T. Engle (jtengle@indiana.edu) Ying Feng (yingfeng@indiana.edu) Robet L. Goldstone (goldsto@indiana.edu) Indiana Univesity Bloomington, IN. 47405 USA Abstact

More information

Physics 235 Chapter 5. Chapter 5 Gravitation

Physics 235 Chapter 5. Chapter 5 Gravitation Chapte 5 Gavitation In this Chapte we will eview the popeties of the gavitational foce. The gavitational foce has been discussed in geat detail in you intoductoy physics couses, and we will pimaily focus

More information

Memory-Aware Sizing for In-Memory Databases

Memory-Aware Sizing for In-Memory Databases Memoy-Awae Sizing fo In-Memoy Databases Kasten Molka, Giuliano Casale, Thomas Molka, Laua Mooe Depatment of Computing, Impeial College London, United Kingdom {k.molka3, g.casale}@impeial.ac.uk SAP HANA

More information

Multiband Microstrip Patch Antenna for Microwave Applications

Multiband Microstrip Patch Antenna for Microwave Applications IOSR Jounal of Electonics and Communication Engineeing (IOSR-JECE) ISSN: 2278-2834, ISBN: 2278-8735. Volume 3, Issue 5 (Sep. - Oct. 2012), PP 43-48 Multiband Micostip Patch Antenna fo Micowave Applications

More information

Electricity transmission network optimization model of supply and demand the case in Taiwan electricity transmission system

Electricity transmission network optimization model of supply and demand the case in Taiwan electricity transmission system Electicity tansmission netwok optimization model of supply and demand the case in Taiwan electicity tansmission system Miao-Sheng Chen a Chien-Liang Wang b,c, Sheng-Chuan Wang d,e a Taichung Banch Gaduate

More information

The Incidence of Social Security Taxes in Economies with Partial. Compliance: Evidence from the SS Reform in Mexico

The Incidence of Social Security Taxes in Economies with Partial. Compliance: Evidence from the SS Reform in Mexico The Incidence of Social Secuity Taxes in Economies ith Patial Compliance: Evidence fom the SS Refom in Mexico Gecia M. Maufo Abstact Looking at impovements in social secuity benefits in Mexico, this pape

More information

Promised Lead-Time Contracts Under Asymmetric Information

Promised Lead-Time Contracts Under Asymmetric Information OPERATIONS RESEARCH Vol. 56, No. 4, July August 28, pp. 898 915 issn 3-364X eissn 1526-5463 8 564 898 infoms doi 1.1287/ope.18.514 28 INFORMS Pomised Lead-Time Contacts Unde Asymmetic Infomation Holly

More information

Semipartial (Part) and Partial Correlation

Semipartial (Part) and Partial Correlation Semipatial (Pat) and Patial Coelation his discussion boows heavily fom Applied Multiple egession/coelation Analysis fo the Behavioal Sciences, by Jacob and Paticia Cohen (975 edition; thee is also an updated

More information

The Impacts of Congestion on Commercial Vehicle Tours

The Impacts of Congestion on Commercial Vehicle Tours Figliozzi 1 The Impacts of Congestion on Commecial Vehicle Tous Miguel Andes Figliozzi Potland State Univesity Maseeh College of Engineeing and Compute Science figliozzi@pdx.edu 5124 wods + 7 Tables +

More information

Unveiling the MPLS Structure on Internet Topology

Unveiling the MPLS Structure on Internet Topology Unveiling the MPLS Stuctue on Intenet Topology Gabiel Davila Revelo, Mauicio Andeson Ricci, Benoit Donnet, José Ignacio Alvaez-Hamelin INTECIN, Facultad de Ingenieía, Univesidad de Buenos Aies Agentina

More information

Over-encryption: Management of Access Control Evolution on Outsourced Data

Over-encryption: Management of Access Control Evolution on Outsourced Data Ove-encyption: Management of Access Contol Evolution on Outsouced Data Sabina De Capitani di Vimecati DTI - Univesità di Milano 26013 Cema - Italy decapita@dti.unimi.it Stefano Paaboschi DIIMM - Univesità

More information

An Energy Efficient and Accurate Slot Synchronization Scheme for Wireless Sensor Networks

An Energy Efficient and Accurate Slot Synchronization Scheme for Wireless Sensor Networks An Enegy Efficient and Accuate Slot Synchonization Scheme fo Wieless Senso Netwoks Lillian Dai Pithwish asu Jason Redi N Technologies, 0 Moulton St., Cambidge, MA 038 ldai@bbn.com pbasu@bbn.com edi@bbn.com

More information

ENABLING INFORMATION GATHERING PATTERNS FOR EMERGENCY RESPONSE WITH THE OPENKNOWLEDGE SYSTEM

ENABLING INFORMATION GATHERING PATTERNS FOR EMERGENCY RESPONSE WITH THE OPENKNOWLEDGE SYSTEM Computing and Infomatics, Vol. 29, 2010, 537 555 ENABLING INFORMATION GATHERING PATTERNS FOR EMERGENCY RESPONSE WITH THE OPENKNOWLEDGE SYSTEM Gaia Tecaichi, Veonica Rizzi, Mauizio Machese Depatment of

More information

arxiv:1110.2612v1 [q-fin.st] 12 Oct 2011

arxiv:1110.2612v1 [q-fin.st] 12 Oct 2011 Maket inefficiency identified by both single and multiple cuency tends T.Toká 1, and D. Hováth 1, 1 Sos Reseach a.s., Stojáenská 3, 040 01 Košice, Slovak Republic Abstact axiv:1110.2612v1 [q-fin.st] 12

More information

Towards Automatic Update of Access Control Policy

Towards Automatic Update of Access Control Policy Towads Automatic Update of Access Contol Policy Jinwei Hu, Yan Zhang, and Ruixuan Li Intelligent Systems Laboatoy, School of Computing and Mathematics Univesity of Westen Sydney, Sydney 1797, Austalia

More information

Converting knowledge Into Practice

Converting knowledge Into Practice Conveting knowledge Into Pactice Boke Nightmae srs Tend Ride By Vladimi Ribakov Ceato of Pips Caie 20 of June 2010 2 0 1 0 C o p y i g h t s V l a d i m i R i b a k o v 1 Disclaime and Risk Wanings Tading

More information

Patent renewals and R&D incentives

Patent renewals and R&D incentives RAND Jounal of Economics Vol. 30, No., Summe 999 pp. 97 3 Patent enewals and R&D incentives Fancesca Conelli* and Mak Schankeman** In a model with moal hazad and asymmetic infomation, we show that it can

More information

2. Orbital dynamics and tides

2. Orbital dynamics and tides 2. Obital dynamics and tides 2.1 The two-body poblem This efes to the mutual gavitational inteaction of two bodies. An exact mathematical solution is possible and staightfowad. In the case that one body

More information

The Binomial Distribution

The Binomial Distribution The Binomial Distibution A. It would be vey tedious if, evey time we had a slightly diffeent poblem, we had to detemine the pobability distibutions fom scatch. Luckily, thee ae enough similaities between

More information

Application of the VISEVA demand generation software to Berlin using publicly available behavioral data

Application of the VISEVA demand generation software to Berlin using publicly available behavioral data Justen, Beuck, Nagel 1 Application of the VISEVA demand geneation softwae to Belin using publicly available behavioal data Submission date: 15-Nov-06 Wods: 5973 Figues and tables: 6 ( = 1500 Wods) Total:

More information

Nice Resource Reservations in Linux

Nice Resource Reservations in Linux Nice Resouce Resevations in Linux Matin Ohlin Depatment of Automatic Contol Lund Univesity Box 118, SE-221 Lund, Sweden Email: Matin.Ohlin@contol.lth.se Matin A. Kjæ Depatment of Automatic Contol Lund

More information

9:6.4 Sample Questions/Requests for Managing Underwriter Candidates

9:6.4 Sample Questions/Requests for Managing Underwriter Candidates 9:6.4 INITIAL PUBLIC OFFERINGS 9:6.4 Sample Questions/Requests fo Managing Undewite Candidates Recent IPO Expeience Please povide a list of all completed o withdawn IPOs in which you fim has paticipated

More information

Trading Volume and Serial Correlation in Stock Returns in Pakistan. Abstract

Trading Volume and Serial Correlation in Stock Returns in Pakistan. Abstract Tading Volume and Seial Coelation in Stock Retuns in Pakistan Khalid Mustafa Assistant Pofesso Depatment of Economics, Univesity of Kaachi e-mail: khalidku@yahoo.com and Mohammed Nishat Pofesso and Chaiman,

More information

Epdf Sulf petroleum, Eflecti and Eeflecti

Epdf Sulf petroleum, Eflecti and Eeflecti ANALYSIS OF GLOBAL WARMING MITIGATION BY WHITE REFLECTING SURFACES Fedeico Rossi, Andea Nicolini Univesity of Peugia, CIRIAF Via G.Duanti 67 0615 Peugia, Italy T: +9-075-585846; F: +9-075-5848470; E: fossi@unipg.it

More information

Alarm transmission through Radio and GSM networks

Alarm transmission through Radio and GSM networks Alam tansmission though Radio and GSM netwoks 2015 Alam tansmission though Radio netwok RR-IP12 RL10 E10C E10C LAN RL1 0 R11 T10 (T10U) Windows MONAS MS NETWORK MCI > GNH > GND > +E > DATA POWER DATA BUS

More information