Online Procurement Auctions for Resource Pooling in Client-Assisted Cloud Storage Systems

Size: px
Start display at page:

Download "Online Procurement Auctions for Resource Pooling in Client-Assisted Cloud Storage Systems"

Transcription

1 Onlne Procurement Auctons for Resource Poolng n Clent-Asssted Cloud Storage Systems Jan Zhao, Xaowen Chu, Ha Lu, Yu-Wng Leung Department of Computer Scence Hong Kong Baptst Unversty Emal: {janzhao, chxw, hlu, Zongpeng L Department of Computer Scence Unversty of Calgary Emal: Abstract Latest developments n cloud computng technologes have enabled a plethora of cloud based data storage servces. Cloud storage servce provders are facng sgnfcant bandwdth cost as the user populaton scales. Such bandwdth cost can be substantally slashed by explorng a hybrd cloud storage archtecture that takes advantage of under-utlzed storage and network resources at storage clents. A crtcal component n the new hybrd cloud storage archtecture s an economc mechansm that ncentvzes clents to contrbute ther local resources, whle at the same tme mnmzes the provder s cost for poolng those resources. Ths work studes onlne procurement aucton mechansms towards these goals. The onlne nature of the aucton s n lne wth asynchronous user request arrvals n practce. After carefully characterzng truthfulness condtons under the onlne procurement aucton paradgm, we prove that truthfulness can be guaranteed by a prce-based allocaton rule and payment rule. Our truthfulness characterzaton actually converts the mechansm desgn problem nto an onlne algorthm desgn problem, wth a margnal prcng functon for resources as varables set by cloud storage servce provders for onlne procurement aucton. We derve the margnal prcng functon for the onlne algorthm. We also prove the compettve rato of the socal cost of our algorthm aganst that of the offlne VCG mechansm and of the resource poolng cost of our algorthm aganst that of the offlne optmal aucton. Smulaton studes drven by real-world traces are conducted to show the effcacy of our onlne aucton mechansm. I. INTRODUCTION Cloud storage servce (e.g., Amazon S3 [1], Dropbox [], Google Drve [3], SkyDrve [4]) has ganed wdespread recognton and adopton by Internet users, many of whom now routnely execute onlne storage and onlne backup tasks over the cloud. Cloud storage provders are facng two natural challenges. Frst, the storage servce hghly depends on centralzed datacenters n the cloud. Accdents and natural dsasters (fre, earthquake, power outage) wll have sgnfcant mpact on the servce performance. Second, the cloud storage provders face sgnfcant cost n purchasng or rentng real estate, cloud nfrastructures, hardware, power and network bandwdth. To address these two challenges, the clent-asssted (or userasssted) cloud storage paradgm s a promsng choce to guarantee hgh data avalablty and relablty whle lowerng the operatonal cost [5]. The clent-asssted desgn s a complementary technology for cloud storage servce that are based solely on servers n datacenters. Recent examples of clent-asssted cloud storage archtecture nclude FSYou [6], AmazngStore [7], Wuala [8] [9], Trton [1], and Symform [11], whch have examned the techncal feasblty of the paradgm of clent-asssted cloud storage servces. However, ths lne of work mostly gnore the need of an ncentve mechansm for users to contrbute ther resources, and assume that unutlzed resources n storage clents can be used for clent-asssted cloud storage archtectures by default, or use an unattractable one by provdng more cloud storage space for clents who contrbute more local storage. The realty s perhaps less optmstc, as autonomous and selfsh users do not have nner-ncentve for contrbutng ther resources (e.g., casual mode n Wuala [8]). Experence from prvate BtTorrent communtes shows that a well desgned ncentve mechansm can motvate users to contrbute ther resources and sgnfcantly mprove the overall system capacty and performance [1] [13]. To help fully make clent-asssted cloud storage a realty, t s mperatve to desgn approprate ncentve mechansms for users to engage. Onlne procurement aucton s a natural canddate that can act as the fnancal catalyst for resource poolng transactons. The procurement form of the aucton s due to the mult-seller (cloud users) one-buyer (storage servce provder) nature of the aucton, and the onlne property s n lne wth asynchronous arrvals of user bds and requests. The goals nclude not only optmzed resource utlzaton across the network, but also fundamental changes n the ecosystem of cloud storage servces leadng to a wn-wn strategy for both sdes. Users can receve monetary rewards for contrbutng ther otherwse dlng resources, whle cloud storage provders can cut ther cost substantally. The resource poolng market s modeled as an onlne procurement aucton market: cloud storage users are economcally motvated to contrbute ther local resources to the storage pool. The storage pool constructed by cloud users can help to reduce the network traffc as well as storage burden at servers n datacenters. Cloud storage users are ncentvzed to submt sellng bds to the cloud storage provder, ndcatng the amount of resources t plans to contrbute, the tme wndow when they are avalable, and the desred remuneraton. Upon recevng a bd, the storage servce provder makes a decson on the amount of resources to be procured, based on a prcng functon. Wth the onlne procurement aucton model for resource poolng market, we face the followng two questons: 1) How should the mechansm guarantee the

2 truthfulness property n such an onlne aucton? ) How well does our mechansm perform when compared wth an offlne Vckrey-Clarke-Groves (VCG) aucton or an offlne optmal aucton? Our general soluton framework s based on convertng the onlne procurement aucton desgn problem nto an onlne algorthm desgn problem. Ths converson s based on constrants that provde condtons characterzng the onlne auctons truthfulness. Hence, we frst characterze the truthfulness for an onlne aucton startng from Myerson s prncple of truthfulness. We defne the allocaton monotoncty for onlne procurement auctons, and prove the exstence of a crtcal payment scheme to guarantee truthfulness n two dmensons, resource avalablty and resource margnal cost under the correspondng monotone allocaton rule. For convenent applcatons n onlne aucton, we prove the equvalence of a prcebased allocaton and payment rule n guaranteeng truthfulness. Wth the prce-based allocaton rule and payment rule, we convert the onlne procurement aucton desgn problem nto an onlne algorthm desgn problem wth margnal prcng functons as varables. An ntegral equaton s derved to solve the margnal prcng functon for guaranteeng a target compettve rato of γ n terms of the socal cost of our algorthm aganst that of the offlne VCG mechansm, and n terms of the resource poolng cost of our algorthm aganst that of the offlne optmal aucton, and dscuss the relaton between the optmal target compettve rato γ and resource pool capacty, number of bds. We summarze our contrbutons as follows. We prove the equvalence of a prce-based allocaton rule and payment rule wth those derved from Myerson s classc prncple of truthfulness, for guaranteeng truthfulness n our onlne aucton. Wth the margnal-prce based allocaton and payment rule, we convert an onlne procurement aucton desgn nto an onlne algorthm desgn problem. We derve an ntegral equaton of the margnal prcng functon that can guarantee our mechansm s performance wth a target compettve rato γ of the socal cost of our algorthm aganst that of the offlne VCG mechansm, and of the resource poolng cost of our algorthm aganst that of the offlne optmal aucton, and yeld the range for the optmal target compettve rato γ. The rest of the paper s organzed as follows. Sec. II presents related work. Sec. III descrbes the system model for clentasssted cloud storage systems and the onlne procurement aucton for resource poolng. Sec. IV characterzes truthfulness for our onlne auctons. Sec. V presents the onlne algorthm desgn problem converted from the onlne aucton problem, and proves the compettve rato acheved by our algorthm. Sec. VI s performance evaluaton of our algorthm under dynamc onlne storage clents extracted from real-world traces. Sec. VII concludes the paper. II. RELATED WORK Clent-asssted cloud storage paradgm desgn motvates much research work [6] [7] [8] [1]. Sun et al. [6] are the frst to desgn, mplement, and deploy a peer-asssted sem-persstent onlne storage system wth clent storage and bandwdth assstance. They make use of unstructured PP overlay constructon, sequental block schedulng mechansm, and server strateges. The desgn objectve s to acheve a tradeoff between fle avalablty and server cost. Yang et al. [7] propose a peer-asssted onlne storage archtecture, AmazngStore, for cloud-based nfrastructures. They organzed nodes usng DHT and gave the number of requred replcas for ganng the requred data avalablty. Hgh fracton of data access requests are served by peers so as to reduce the server cost and mprove the data avalablty. Mager et al. [8] conduct a measurement study of Wuala, a popular PP-asssted onlne storage and sharng system. Toka et al. [14] study the mpact of data placement and bandwdth allocaton on the tme requred to complete a backup and restore operaton and the clents costs. Davol et al. [1] propose solutons for acceleraton on data sharng, mprovng the consstency, and reducng latency for agreement protocol preventng concurrent accesses on shared data. Ths group of work proposes PP-asssted onlne storage system desgn or conducts measurement work on such desgns based on an assumpton that peers contrbute ther resources by default. Only Mager et al. [8] and Toka et al. [14] menton the ncentve problem of peers contrbutng local resources. There are two modes for peers n Wuala. One type of peers are casual peers who contrbute lttle storage and the other s storage peers who trade local storage for ncreased cloud storage space or reduced cloud storage cost. In ths paper, we propose an onlne procurement aucton for tradng peers local resource, whch serves as an ncentve mechansm for peers to contrbute ther local resources. Aucton mechansms have been studed by some researchers [15] [16] and been used n some scenaros such as PP streamng [17], WF prcng [18], spectrum auctons [19] [], cloud computng prcng [1] [] [3]. Lav et al. [15] present compettve analyss for truthful onlne auctons. Ther method s based on a threat-based approach proposed by Yanv et al. [4]. Our compettve analyss s based on dervng and solvng an ntegral equaton, whch s qute dfferent from thers. Hajaghay et al. [16] study the onlne auctons for reusable goods. They propose smlar truthfulness condtons for onlne auctons, but they do not form an onlne algorthm desgn framework for the onlne aucton mechansm desgn. Fredman et al. [18] consder extendng the standard results of offlne mechansm desgn to apply to mechansm desgn for onlne problems such as WF Prcng for users arrvng over tme. Deek et al. [] study the onlne aucton for spectrum allocaton. They propose a 3D bn-packng based allocaton and tme-smoothed crtcal value based prcng scheme, and evaluate ts performance through experments. Zhang et al. [1] and Sh et al. [] study the onlne aucton for cloud computng resource allocaton. Zhang et al. [1] propose the

3 3 condtons of payment to ensure the truthfulness. Our method propose the allocaton monotoncty, whch s the necessary and suffcent condton to guarantee the truthfulness. Sh et al. [] apply the threat-based approach for the compettve analyss. Our onlne procurement aucton desgn framework can be seen as an optmal onlne aucton desgn, whch expands Myerson s framework on optmal aucton desgn [5]. III. MODEL FORMULATION In a clent-asssted cloud storage system, the storage servce provder ams to procure ts clents unused resources to complment server resources n ts own datacenters. Such a hybrd, dstrbuted storage archtecture helps acheve hgher storage cost effcency and robustness aganst sngle pont of falure. The cloud servce provder needs storage and network bandwdth from clents to deploy such clent-asssted storage servce. Resource s assumed as a combnaton of storage and network bandwdth at a calculated rato for fle avalablty and accessblty. Let S be the resource capacty that the servce provder ams to procure durng tme [, T ]. In the onlne procurement aucton mechansm M, the cloud storage provder acts as the auctoneer. Storage clents on the Internet are bdders who may sell ther own resources to the auctoneer. Clents are dynamc n that they may arrve and depart at any tme. The set of bdders s unknown to the auctoneer a pror, and a bdder learns ts type at the tme of bddng. Consequently, the storage provder receves a stream of bds, B = (b 1, b,..., b j,...), from dynamc clents, and needs to make decsons on each bd upon ts submsson. The bd specfes the avalable resource the bdder can contrbute and the monetary remuneraton asked n return. More specfcally, the j-th receved bd s a tuple b j = (â j, ˆd j, ˆQ j, ĉ j ), where â j s announced arrval tme of avalable resources, ˆdj s announced departure tme of avalable resources, ˆQj s announced avalable resource capacty, and ĉ j s announced margnal cost, whch s the cost for bdder to provde one more unt of resource for one tme unt. The type of the bdder submttng bd b j (referred to as bdder j) s denoted by v j = (a j, d j, Q j, c j ), whch s prvate nformaton known to bdder j only. Let B j denote all other bds except the j-th bd. When the auctoneer receves bd b j, t determnes the resource capacty to procure, q j (t, b j, B j ), q j ˆQ j durng tme t [â j, ˆd j ], and the correspondng payment, p j (t, b j, B j ). We call (q j, p j ) the allocaton rule and payment rule of the onlne procurement aucton. We wll use q j (t, b j ), p j (t, b j ) nstead of q j (t, b j, B j ), p j (t, b j, B j ) when there s no confuson. Bdder j s total utlty s the total payment t receves mnus ts total cost to offer the procured resources, ntegrated over the tme wndow [â j, ˆd j ]: U j (q j, p j ) = ˆd j â j u j (t, q j, p j )dt, where u j(t, q j, p j) = p j(t, b j) c jq j(t, b j). (1) Eqn. (1) ensures ndvdual ratonalty of bdders. We are nterested n mechansms wth the truthfulness property. We frst examne the bdders strategy space. For a j, d j, Q j, bdders can not report an earler arrval tme â j < a j, a later departure tme ˆd j > d j, a capacty larger than what s avalable ˆQ j > Q j, as such false reports are easly detected. Hence, the bdders strategy space s â j a j, ˆdj d j, ˆQ j Q j, ĉ j c j. The defnton of truthfulness for onlne aucton mechansms s as follows. Defnton 1 (Truthfulness). An onlne aucton mechansm M s truthful, f for any bdder j, regardless of the type reports of other bdders,.e., past and future bds, declarng a bd that reveals ts true type can maxmze ts utlty. I.e., for a bdder wth type v j = (a j, d j, Q j, c j ), every bd b j = (â j, ˆd j, ˆQ j, ĉ j ) of the bdder satsfyng â j a j, ˆdj d j, ˆQj Q j, ĉ j c j, we have U(q j (t, v j ), p j (t, v j )) U(q j (t, b j ), p j (t, b j )). () The defnton of truthfulness for our model s two-fold: one s to reveal resource avalablty,.e., a j, d j, Q j, truthfully; the other s to reveal resource cost,.e., c j, truthfully. Let B t be the set of stream bds receved by tme t, B t = {b j â j t}. Wth B t, we can defne the servce provder s completeness rato for any tme t s resource poolng after recevng streamng bds B t, R(t, t), R(t j B, t) = t q j(t, b j) [, 1], t t T. (3) S When the requred capacty S s not acheved by tme t through clent-asssted resources at tme t,.e., R(t, t) < 1, t wll need to use the storage and bandwdth from servers n the datacenters to compensate. Let c s be the margnal resource cost from servers. The cost of the cloud provder for provdng a capacty of S resource under a bddng sequence B under mechansm M : (q j, p j ) s denoted by C M (B). The resource poolng cost s the total payment he pays for resource poolng plus the server cost for compensatng the remanng resource. C M(B) = j B ˆdj â j p j(t, b j)dt + c s S[1 R(t, t)]dt. (4) The onlne procurement aucton mechansm desgn problem can be modeled as an onlne optmzaton problem as follows, mn C M(B) s.t. (1)()(3) The objectve functon C M (B) can be rearranged as follows: C M(B) = [ j B t p j(t, b j) + c s S[1 R(t, t)]]dt And the constrants about the allocaton rule and payment rule are also decoupled at dfferent tme ponts. The optmzaton problem s equvalent to mnmzng the followng problem for any tme t [, T ], mn p j(t, b j) + c s S[1 R(t, t)] j B t s.t. (1)() R(t, t) [, 1]. (5) Optmzaton problem (5) can be seen as a sub-problem of resource poolng for tme t [, T ]. Table I summarzes mportant notatons for ease of reference.

4 4 TABLE I: Important Notatons S resource pool capacty. T tme span of resource. W M (B) the total socal cost. B stream of bds. M the onlne procurement aucton mechansm. B j all other bds except the j-th bd. b j the j-th bd n the bd stream. v j true type of the j-th bd. a j true arrval tme of bdder j s resource. d j true departure tme of bdder j s resource. Q j true avalable resource of bdder j. c j true unt resource cost for unt tme of bdder j. q j (t, b j ) resource capacty procured from bdder j. p j (t, b j ) payment to bdder j for tme t s resource. U j (q j, p j ) bdder j s utlty. B t streamng bds receved by tme t. R(t, t) completeness rato of tme t s resource poolng based on procurement by t. c s unt resource cost for unt tme of storage provder. C M (B) cloud storage provder s total cost. IV. TRUTHFULNESS CHARACTERIZATION FOR ONLINE AUCTIONS A. Revstng Myerson s Prncple of Truthfulness Myerson [5] formulated the classc characterzaton of offlne truthful mechansms n a general Bayesan settng, for auctons of a sngle ndvsble good. Lemma 1 (Myerson, 1981). Let P (b ) be the probablty of bdder wth bd b wnnng an aucton. A mechansm s truthful f and only f the followngs hold for a fxed b : P (b ) s monotoncally non-decreasng n b ; Bdder bddng b s charged b P (b ) b P (b)db Gopnathan et al. [6] ponted out the two nterpretatons of Myerson s prncple: () there exsts a mnmum bd b such that wll wn only f t bds at least b, and () the payment charged to for a fxed b s ndependent of b. We wll frst present the truthfulness characterzaton for our onlne procurement aucton by adaptng and extendng Myerson s prncple of truthfulness. In our onlne procurement aucton, to guarantee truthfulness s to guarantee that bdders bd ther true types n both resource avalablty and resource margnal cost,.e., b j = v j (â j = a j, ˆd j = d j, ˆQ j = Q j, ĉ j = c j ), regardless of past and future streams of other bds. Myerson s prncple of truthfulness frst ponts out the monotoncty crteron of allocaton rule P (b ) under a Bayesan game model for auctons of a sngle ndvsble good. Our onlne procurement aucton s an onlne reverse aucton for a dvsble good wth quantty S. Hence, we frst defne a correspondng allocaton monotoncty for our aucton, whch s necessary and suffcent for the exstence of a payment rule elctng truthful bds. Defnton (Allocaton Monotoncty). Consder two bds b j = (a j, d j, Q j, c j ) and b j = (â j, ˆd j, ˆQ j, ĉ j ) n the onlne procurement aucton. If â j a j, ˆd j d j, ˆQj Q j, ĉ j c j,.e., bd b j s resource avalablty s a subset of that of bd b j, and bd b j s resource margnal cost s no smaller than that of bd b j, we say bd b j domnates bd b j, denoted by b j b j. An allocaton rule q s monotone f ˆdj â j q j(t, b j)dt d j a j q j(t, b j)dt j, f b j b j. (6) Let us examne the dfference between the allocaton monotoncty crteron between Myerson s prncple and our defnton. Myerson consders a bayesan game model for auctons of one sngle ndvsble good. Hence, the allocaton probablty to one bdder when havng ts bd b, P (b ), s used to represent the allocaton rule. Ths allocaton probablty s used because we consder other bds satsfy a probablty dstrbuton and the probablty dstrbuton of other bds s a pror. In our onlne procurement aucton for dvsble good wth quantty S, we use the allocaton quantty to one bdder when havng ts bd b j, ˆd j â j q j (t, b j )dt, to represent the allocaton rule, gven other bds are fxed. When other bds vary accordng to a pror probablty dstrbuton, the expectaton of the allocaton quanttes s stll monotone wth b j. Ths s consstent wth P (b ) s monotoncty n a sngle ndvsble good s aucton. Gven allocaton monotoncty, we present the payment scheme to mplement a truthful onlne procurement aucton n Theorem 1. Theorem 1. There s a payment rule p such that the mechansm M : (q, p) s truthful f and only f the allocaton rule q s monotone. And the payment scheme can be expressed as follows, p j(t, b j) = ĉ j q j(t, b j) + ĉ j q j(t, (â j, ˆd j, ˆQ j, x))dx. (7) Theorem 1 can be seen as an extenson of Myerson s prncple of truthfulness to onlne procurement auctons of a dvsble good. We prove the truthfulness of resource avalablty and resource margnal cost under condtons n Theorem 1. Detals of ts proof are n Appendx IX-A. B. A Prce-based Allocaton and Payment Rule Theorem 1 s characterzaton of truthfulness stands n the vewpont of the allocaton rule. Once the allocaton rule s determned, the payments are correspondngly determned. Ths characterzaton s convenent for offlne auctons where all bds are collected smultaneously and then the allocaton decsons for all related bds are made at the same tme. In such cases, the allocaton rule s obvous. However, ths does not apply to onlne auctons where bds arrve at dfferent tmes, wth allocaton decsons made at realtme. We wll present an equvalent condton to guarantee truthfulness of onlne procurement aucton mechansms, n whch a margnal prcng functon wll be ntroduced. The allocaton and payment are based on the prcng functon. Prce-based Allocaton and Payment: Let Ψ(t, R) be the margnal prcng functon of the cloud storage provder for procurng tme t s resources from clents when the completeness rato of tme t s resource poolng s R. It s easy to see that Ψ(t, R) s bd-ndependent.

5 5 In an onlne procurement aucton for S resource capacty, when the cloud storage provder receves a bd b j = (â j, ˆd j, ˆQ j, ĉ j ), t procures q j (t, b j ) resource from the bdder and pays p j (t, b j ) to the bdder. q j (t, b j ) and p j (t, b j ) are determned as follows, q j(t, b j) =, f ĉ j Ψ(t, R); mn{ ˆQ j, S [Ψ 1 (t, ĉ j) R(t, â j )]}, f ĉ j < Ψ(t, R). p j(t, b j) = R(t,âj ) (8) R(t,â j ) S Ψ(t, R)dR (9) Here, R(t, â j ) = q k Bâ k (t, b k )/S s the completeness j rato for tme t s resource poolng, just before recevng bd b j. We have Theorem for the truthfulness of prce-based onlne procurement auctons. Theorem. When a cloud storage provder determnes ts allocaton and payment accordng to Eqn. (8) and (9) when recevng a bd, where Ψ(t, R) s non-ncreasng n completeness rato R, the mechansm M : (q, p) s truthful. Proof: We frst prove monotoncty of the allocaton rule. Gven two bds b = (, ˆd, ˆQ, ĉ ), b = (a, d, Q, c ), b b from a bdder, we wll prove q (t, b ) q (t, b ). Let us examne the allocaton under a new bd b = (a, d, ˆQ, ĉ ). It s easy to verfy that b b b. As the resource avalablty of b domnates b, we have a. As bd b arrves later, the completeness rato of tme t s resource poolng before recevng bd b s smaller than or equal to that before recevng bd b,.e., S R(t, â ) = k Bâ q k (t, b k ) k B a q k (t, b k ) = S R(t, a ) The avalable resource and resource margnal cost s the same for b and b. Accordng to Eqn. 8, we have q (t, b ) q (t, b ). As for bd b and b, the resource arrval tmes are the same for the two bds. Hence, the completeness rato of tme t s resource poolng when recevng bd b or b are the same. As Ψ(t, R) s non-ncreasng wth R, we can verfy that ts nverse functon Ψ 1 (t, c) s also non-ncreasng wth c. Because ĉ c, we have Ψ 1 (t, ĉ ) Ψ 1 (t, c ) We also have ˆQ Q, ths results n q (t, b ) q (t, b ) accordng to Eqn. (8). In concluson, we proved q (t, b ) q (t, b ). The allocaton rule based on the margnal prcng functon Ψ(t, R)s monotone. Wth the monotone allocaton rule, based on Theorem 1, we know that the payment rule n Eqn. (7) truthfully mplements the allocaton rule. The prce-based payment Eqn. (9) can be derved by rearrangng Eqn. (7) based on the prcebased allocaton Eqn. (8). V. A COMPETITIVE ONLINE PROCUREMENT ALGORITHM The prce-based allocaton rule and payment rule work n concert to not only guarantee the truthfulness of an onlne aucton mechansm, but also help us convert an onlne aucton desgn problem nto an onlne algorthm desgn problem. The onlne algorthm desgn problem has the margnal prcng functon as ts varable. Usng an offlne VCG aucton and an offlne optmal aucton as benchmarks for socal cost and cloud storage provder s poolng cost respectvely, we derve an ntegral equaton for guaranteeng a target compettve rato γ, and further derve the margnal prcng functon from the ntegral equaton. A. Onlne Algorthm Desgn Problem Converted from Onlne Aucton Desgn By substtutng the prce-based payment rule (9) nto the optmzaton problem (5), we obtan the followng onlne algorthm desgn problem, mn s.t. j B t R(t,âj ) R(t,â j ) Ψ(t, R)S dr + c s S[1 R(t, t)] Ψ(t, R) s a non-ncreasng functon Ψ(t, 1) c mn (1) Condton () of optmzaton problem (5),.e., property of truthfulness, can be guaranteed based on the prce-based allocaton and payment rule. Ψ(t, 1) c mn guarantees condton R(t, t) [, 1] of (5). Condton (1) s satsfed as the prce-based allocaton rule and payment rule only procure resource from clents wth resource margnal cost for tme t smaller than Ψ(t, R), and pay clents prces that are hgher than the resource margnal cost. Hence, for any bdder j, U j (q j, p j ). We summarze our algorthm framework for the onlne procurement aucton n Algorthm 1. B. Margnal Prcng Functon Dervaton and Compettve Rato Analyss To mplement Algorthm 1, we need to gve the margnal prcng functon. We wll see the prcng functon wll affect the compettve rato of our onlne procurement aucton. In our onlne procurement aucton desgn, we am at mnmzng the cloud provder s resource poolng cost, whch s the summaton of the payment of the cloud storage provder for procurng resource from clents and the cost of provdng complmentary resource from servers. Wth the socal cost of offlne VCG aucton as a lower bound of the resource poolng cost of the offlne optmal aucton, we analyze the compettve rato of the resource poolng cost aganst that of the offlne optmal aucton. Besdes the resource poolng cost, another mportant metrc s the socal cost,.e., the total cost for clents and servers to provde S resource. We benchmark our onlne algorthm aganst the offlne VCG aucton that acheves the mnmum socal cost.

6 6 Algorthm 1 Cloud Storage Provder s Algorthm for the Onlne Procurement Aucton Input: Margnal prcng functon Ψ(t, R), S, T. Output: q j (t, b j ), p j (t, b j ) 1: Intalze the completeness rato of tme t [, T ] s resource poolng R t = : whle Receve a bd b j do 3: for t [â j, ˆd j ] do 4: f R t = 1 then 5: Contnue 6: end f 7: f ĉ j < Ψ(t, R t ) then 8: f Ψ 1 (t, ĉ j ) > 1 then 9: Set Ψ 1 (t, ĉ j ) = 1 1: end f 11: Procure q j (t, b j ) resource from bdder j accordng to Eqn.(8) 1: Pay p j (t, b j ) to bdder j accordng to Eqn.(9) 13: Update R t = R t + qj(t,bj) S 14: end f 15: end for 16: end whle The socal cost for provdng tme t s S resource under our onlne algorthm s W M(t, B) = j B t c jq j(t, b j) + c s S[1 R(t, t)]. (11) The total socal cost s the ntegraton of W M (t, B) among tme [, T ],.e., W M (B) = W M(t, B)dt. Theorem 3. When the cloud storage provder uses the margnal prcng functon as Ψ(t, R) = c s c s (1 1 γ )e R γ, where R s the completeness rato of procured resource for tme t, our prce-based onlne procurement aucton wll acheve the resource poolng cost no larger than γ tmes that of an offlne optmal aucton,.e., C M (B) γ C opt (B) and acheve the socal cost no larger than γ tmes that of the offlne VCG aucton,.e., W M (B) γ W vcg (B). The optmal γ should be among the range of [γ,mn, γ,max ]. Here γ,mn and γ,max are the solutons to equatons (1 1 γ ) e 1 γ = 1 cmax c s, (1 1 γ ) e 1 γ = 1 c mn c s respectvely. C opt (B), W vcg (B) s the cloud storage provder s resource poolng cost under offlne optmal aucton and the socal cost under offlne VCG aucton respectvely. Proof: We frst check the socal cost under our onlne procurement aucton and offlne VCG aucton. Gven some bddng sequence B, let q j (t, b j ) be the quantty procured from bdder j for poolng tme t s resource, and Ψ(t, R(t, â j )) the lowest procurement prce for poolng tme t s unt resource of unt tme from bdder j. Let l be the last bdder wth resource procured, the correspondng lowest procurement prce for poolng tme t s resource from bdder l s Ψ(t, R(t, â l )). Let R(t, â l ) be the fnal completeness rato for tme t s resource procurement after procurng bdder l s resource. The poolng cost for tme t s resource s C M(t, B) = R(t,âl ) Ψ(t, R)S dr + c s [S S R(t, â l )] When the cloud provder procures resource, t pays more than the bdders cost. Hence, the socal cost for provdng S resource wll be no larger than the cloud provder s resource poolng cost,.e., W M (t, B) C M (t, B). An offlne aucton mechansm can collect all bds B = (b 1, b,..., b j,...), then make allocaton and payment decsons. As n our procurement aucton, servers resource can be utlzed as a reserved backup resource. Correspondngly, we add a new bd b wth announced cost c s and unbounded resource capacty. For procurng tme t s resource, let (b (vcg) 1, b (vcg), b (vcg) 3,..., b (vcg) x ) be the ascendng sequence of procured bds accordng to the announced cost under offlne VCG aucton. The prce pad to bdders s no smaller than the announced cost of bd b (vcg) x. As the offlne VCG aucton procure the total S capacty, ths means the total resource capacty of the bds procured n offlne VCG aucton s no smaller than the procured resource capacty from clents n our onlne procurement aucton. Hence, the announced cost of b vcg x s larger than or equal to Ψ(t, R(t, â l )). Otherwse, all bds n (b (vcg) 1, b (vcg), b (vcg) 3,..., b (vcg) x ) have announced cost smaller than Ψ(t, R(t, â l )). our onlne algorthm wll procure more than S resource capacty. Hence, the socal cost under offlne VCG aucton for provdng tme t s resource s W vcg (t, B) S c (vcg) x cost of bd b (vcg) S Ψ(t, R(t, â l )). Here c (vcg) x x. s the announced Wth the socal cost for provdng tme t s resource under our onlne procurement aucton and offlne VCG aucton, the compettve rato n terms of the total socal cost can be calculated as follows, W M(B) W = WM(t, B) vcg(b) T Wvcg(t, B) [ R(t,â l ) Ψ(t, R)S dr + c s (S S R(t, â l ))]dt [S Ψ(t, R(t, â l))]dt To guarantee a compettve rato of γ, we can let R(t,âl ) Ψ(t, R)S dr + c s (S S R(t, â l )) = γ (1) S Ψ(t, R(t, â l )) By solvng the ntegral equaton (1), we can derve the expresson of Ψ(t, R) as: Ψ(t, R) = c s c s(1 1 γ ) e R γ Let us consder the bound condton of the margnal prcng functon: () If Ψ(t, 1) < c mn, ths means Ψ(t, R) wll becomes equal to c mn at a pont R < 1, and the cloud storage provder wll stop procurng resources from bdders as the completeness rato does not reach 1. Ths may waste the chances to procure resource wth lower announced cost. use lower cost than server cost to buy comng bdders resources. Hence, ths s not optmal for settng target compettve rato, γ.

7 7 () If c max > Ψ(t, 1) > c mn, ths means the margnal prcng functon wll not be contnuous at R = 1. The cloud storage provder needs to set the prce as c mn when R = 1. But the useful scenaros for settng c max > Ψ(t, 1 ) > c mn s when the number of bdders s small enough or target resource pool capacty S s large enough. In such scenaros, R can not reach 1 due to scare resource from storage clents. Hence, the optmal choce s to accept as many bds as possble. The optmal bound condton s settng Ψ(t, 1) = c max. (1 1 γ ) e 1 γ = 1 c max c s. We can get the optmal target compettve 1 rato s γ,mn =, here, W W ( (x) s a Lambert cmax/cs 1 )+1 e W functon and W (x) 1. () If the number of bds s large enough or target resource pool capacty S s small enough, R can easly reach 1, the optmal soluton s settng Ψ(t, 1) = c mn. The optmal target 1 compettve rato γ,max =. W ( c mn /cs 1 e )+1 Summarzng the above analyss, we can see the optmal target compettve rato s n the range γ [γ,mn, γ,max ]. Next, let us see the resource poolng cost under our onlne procurement aucton and offlne optmal aucton. As the cloud storage provder pays more than bdders announced cost for procurng resource from clents, the resource poolng cost under the offlne optmal aucton should be larger than or equal to the socal cost under the offlne VCG aucton,.e., we have C opt (t, B) W vcg (t, B) S Ψ(t, R(t, â l )). Hence, the compettve rato of our onlne procurement aucton, n terms of resource poolng cost, s: C M(B) C = CM(t, B) opt(b) T Copt(t, B) [ R(t,â l ) Ψ(t, R)S dr + c s (S S R(t, â l ))]dt [S Ψ(t, R(t, â l))]dt whch equals the compettve rato γ of socal cost. VI. PERFORMANCE EVALUATIONS In ths secton, we study the socal cost and resource poolng cost acheved by our onlne procurement algorthm n the context of real-world traces. A. Experment Settngs The clents arrval/departure tme and avalable bandwdth are extracted from real-world traces [7]. We use the sesson 9 trace for clent level data collected from transamrt.net wth 5 mnute samplng ntervals, and plot the onlne statstcal nformaton n Fg. 1. By default, the cost of clents unt resource for a unt tme s generated randomly n the range of [.1,.5]$/GB per hour. The cloud storage provder s server cost for provdng resource s.1$/gb per hour. The total resource poolng capacty of the cloud storage provder s 5KB/s. We consder a tme range of 7 days. The number of bds s 7. We compare the socal cost acheved by our onlne procurement wth that of the offlne VCG aucton. We compare the resource poolng cost by our onlne procurement wth that of optmal aucton ndrectly. As the resource poolng cost of offlne optmal aucton should be larger than the socal cost acheved by offlne VCG aucton, we use the socal cost acheved by offlne VCG aucton as a lower bound for t. Hence, n our experments, the socal cost and resource poolng cost of our onlne procurement aucton are both compared wth the socal cost of the offlne VCG aucton. B. Performance of Our Onlne Procurement Mechansm Fg. compares the resource poolng cost and socal cost of our onlne procurement aucton mechansm wth offlne auctons. We make comparsons under dfferent resource poolng capacty, number of bds, and rato of server cost and average bddng cost. We use the target compettve rato γ,max n the margnal prcng functon. Fg. a vares the resource capacty among [1, 1]KB/s. Fg. b vares the number of bds among [1, 7]. Fg. c vares the rato between margnal server cost c s and average bddng cost among [, 6]. Our onlne procurement aucton acheves the socal cost and resource poolng cost wthn the target compettve rato. In most cases, our onlne procurement aucton s resource poolng cost s lower than that of the offlne VCG aucton. C. How Does Performance Change wth γ? Fg. 3 & Fg. 4 show the change of real cost rato wth the target compettve rato. The black star lnes n Fg. 3a & Fg. 3b are the poston of γ,mn. In Fg. 3a, S = 5KB/s, whch means clents resource s enough, we can see as the target compettve rato equals γ,max, our onlne procurement aucton acheves the optmal compettve rato. In Fg. 3b, S = 8KB/s, whch means clents resource s not enough, we can see the real cost rato s not senstve to the target compettve rato. These two fgures verfy the optmal target compettve rato s among [γ,mn, γ,max ]. Fg. 3c, Fg. 3d and Fg. 4 show the real rato for resource poolng (upper one) and the real rato for socal cost (lower one) under dfferent target compettve rato and changng resource pool capacty, number of bds, and rato of server cost and average bddng cost, rato of the maxmum and mnmum bddng cost. We can see when our algorthm acheves the best performance, the target compettve rato s no larger than γ,max. As the target compettve rato becomes small enough, the real cost rato may even become hgher, whch means the target compettve rato should not be too small. As S s large enough, or number of bds s small enough, the real performance s nsenstve to γ. VII. CONCLUSIONS Clent-asssted cloud storage s emergng as a new, exctng hybrd archtecture for onlne storage systems that represent a potental wn-wn soluton for both cloud storage provders and cloud users. Whle recent research along ths thread have focused on techncal challenges, ths work represents the frst effort that examnes the economc sde of the pcture. We desgn an onlne procurement aucton for resource poolng by the provder, whch serves as a fnancal catalyst for brngng

8 8 No. of Peers No. of Peers No. of Peers (a) Arrval Tme Dstrbuton (days) (b) Onlne Tme Dstrbuton (hours) (c) Resource Capacty Dstrbuton (MB/s) Fg. 1: Statstcal Informaton of Clent Onlne Behavor Cost($) W vcg C vcg γ W vcg W onlne C onlne S (KB/s) (a) Cost($) W vcg C vcg γ W vcg W onlne C onlne No. of Bds (b) Cost($) W vcg C vcg γ W vcg W onlne C onlne Server Cost/Average Bddng Cost Fg. : Performance of our onlne procurement aucton under varyng resource poolng capacty S, No. of bds, and rato among server cost and average bddng cost (c) 8 6 Real Rato for Socal Cost Real Rato for Pool Cost Target Compettve Rato 8 6 Real Rato for Socal Cost Real Rato for Pool Cost Target Compettve Rato Cost Rato 4 Cost Rato γ/γ,max (a) Real Cost Rato, S=5KB/s γ/γ,max (b) Real Cost Rato, S=8KB/s (c) Real Cost Rato under Varyng(d) Real Cost Rato under Varyng Poolng Capacty No. of Bds Fg. 3: Real cost rato wth varyng γ clent-asssted cloud storage nto realty. The aucton proposed s truthful and guarantees a provable compettve rato aganst optmal offlne algorthms. VIII. ACKNOWLEDGEMENTS Ths work s supported by Hong Kong GRF grant HKBU 141. We thank the revewers for ther valuable comments. REFERENCES [1] Amazon S3, [] Dropbox, https://www.dropbox.com. [3] Google Drve, https://drve.google.com. [4] OneDrve, https://onedrve.lve.com. [5] X. Chu, H. Lu, Y.-W. Leung, Z. L, and M. Le, User-Asssted Cloud Storage System: Opportuntes and Challenges, IEEE COMSOC MMTC E-Letter, January 13. [6] Y. Sun, F. Lu, B. L, B. L, and X. Zhang, FSYou: Peer-Asssted Sem-Persstent Onlne Storage at a Large Scale, n Proc. of INFOCOM, 9. [7] Z. Yang, B. Zhao, Y. Xng, S. Dng, F. Xao, and Y. Da, AmazngStore: Avalable, Low-cost Onlne Storage Servce Usng Cloudlets, n Proc. of IPTPS, 1. [8] T. Mager, E. Bersack, and P. Mchard, A Measurement Study of the Wuala On-lne Storage Servce, n Proc. of PP Computng, 1. [9] Wuala, https://www.wuala.com/. [1] A.Davol and A.Me, Trton: a Peer-Asssted Cloud Storage System, n Proc. of PaPEC, 14. [11] Symform, [1] X. Chen, X. Chu, and Y. Jang, Measurements, Analyss and Modelng of Prvate Tracker, n Proc. of PP Computng, August 1. [13] X. Chu, X. Chen, A. L. JIa, J. A. Pouwelse, and D. H. J. Epema, Dssectng Darknets: Measurement and Performance Analyss, ACM Trans. Internet Technol., vol. 13, no. 3, pp. 7:1 7:5, May 14. [14] L.Toka, M.Amco, and P.Mchard, Onlne Data Backup: a Peer-

9 9 (a) Real Cost Rato vs. Varyng Rato among Server Cost and Average Bddng Cost (b) Real Cost Rato vs. Varyng Rato among Maxmum Bddng Cost and Mnmum Bddng Cost Fg. 4: Real cost rato under varyng γ and rato among server cost and bddng cost Asssted Approach, n Proc. of PP Computng, 1. [15] R. Lav and N. Nsan, Compettve Analyss of Incentve Compatble On-Lne Auctons, n Proc. of ACM EC,. [16] M.Hajaghay, R. Klenberg, M.Mahdan, and D. Parkes, Onlne Auctons wth Reusable Goods, n Proc. of ACM EC, 5. [17] X. Chu, K. Zhao, Z. L, and A. Mahant, Aucton-Based On-Demand PP Mn-Cost Meda Streamng wth Network Codng, IEEE Trans. Parallel Dstrb. Syst., vol., no. 1, pp , Dec. 9. [18] E. Fredman and D. Parkes, Prcng WF at Starbucks Issues n Onlne Mechansm Desgn, n Proc. of ACM EC, 3. [19] X.-Y. L, P. Xu, S. Tang, and X. Chu, Spectrum Bddng n Wreless Networks and Related, n Proc. of COCOON, 8, pp [] L. Deek, X. Zhou, K. Almeroth, and H. Zheng, To Preempt or Not: Tacklng Bd and Tme-based Cheatng n Onlne Spectrum Auctons, n Proc. of INFOCOM, 11. [1] H. Zhang, B. L, H. Jang, F. Lu, A. Vaslakos, and J. Lu, A Framework for Truthful Onlne Auctons n Cloud Computng wth Heterogeneous User Demands, n Proc. of INFOCOM, 13. [] W. Sh, C. Wu, and Z. L, RSMOA: A Revenue and Socal Welfare Maxmzng Onlne Aucton for Dynamc Cloud Resource Provsonng, n Proc. of IWQOS, 14. [3] H. Fu, Z. L, C. Wu, and X. Chu, Core-Selectng Auctons for Dynamcally Allocatng Heterogeneous VMs n Cloud Computng, n Proc. of IEEE CLOUD, June-July 14. [4] R.E.Yanv, A.Fat, R.Karp, and G.Turpn, Compettve Analyss of Fnancal Games, n Proc. of FOCS, 199. [5] R.B.Myerson, Optmal Aucton Desgn, Mathematcs of Operatons Research, vol. 6, pp , [6] A. Gopnathan, Z. L, and C. Wu, Strategyproof Auctons for Balancng Socal Welfare and Farness n Secondary Spectrum Markets, n Proc. of INFOCOM, 11. [7] B. Zhang, A. Iosup, and D. Epema, The Peer-to-Peer Trace Archve: Desgn and Comparatve Trace Analyss, TU Delft, Tech. Rep., Aprl 1. A. Proof of Theorem 1 IX. APPENDIX Proof: (a) We frst prove the f part. Let q be a monotone allocaton rule and consder a bd b j = (a j, d j, Q j, c j ). We show that the allocaton rule q n combnaton wth the payment rule p n Eqn. 7 consttute a truthful mechansm. We prove the truthfulness for resource avalablty and resource cost respectvely. () We frst prove the truthfulness for resource cost revelaton,.e., bdders bddng ther true cost wll maxmze ther utlty. We prove t by contradcton. If the mechansm s not truthful, there s a bdder, wth a true type v = {a, d, Q, c }, and a non-truthful bd b = (, ˆd, ˆQ, ĉ ), b v such that the utlty U[q (t, b ), p (t, b )] of bdder f t bds b s strctly greater than the utlty U[q (t, v ), p (t, v )], whch can be wrtten as follows, [ ] (ĉ c )q (t, b ) + q (t, (, ˆd, ˆQ, x))dx dt > a c ĉ q (t, (a, d, Q, x))dxdt As the allocaton rule s monotone, and a, ˆd d, ˆQ Q, we have q (t, (a, d, Q, x))dxdt a c c q (t, (, ˆd, ˆQ, x))dxdt Wth the above two nequaltes, we get the followng relatonshp, ĉ (ĉ c )q (t, b )dt > q (t, (, ˆd, ˆQ, x))dxdt Hence, f ĉ > c, by dvdng both sdes by ĉ c we get ˆd q (t, b )dt s strctly larger than the average of q (t, (, ˆd, ˆQ, x))dt over x [c, ĉ ], whch contradcts the monotoncty of q. If ĉ < c, by dvdng both sdes by ĉ c, we get ˆd q (t, b )dt s strctly less than the average of ˆd q (t, (, ˆd, ˆQ, x))dt over x [ĉ, c ], whch agan contradcts the monotoncty of q. Ths contradcton proves the truthfulness of the mechansm M : (q, p) for resource cost. () Secondly, we prove the truthfulness for resource avalablty. As we have prove the truthfulness for resource cost,.e., for any bd b = (, ˆd, ˆQ, ĉ ), the bd b = (, ˆd, ˆQ, c ) wth the same resource avalablty and the true cost value wll acheve a larger utlty. Next we wll prove the utlty under bd v wll be larger than that under bd b It s obvous that. U[q (t, v ), p (t, v )] = c a c q (t, (a, d, Q, x))dxdt q (t, (, ˆd, ˆQ, x))dxdt = U[q (t, b ), p (t, b )] c Hence, we prove the truthfulness for resource avalablty. (b) We then prove the only f part. We have a truthful mechansm M : (q, p). Consder a bdder and ts two possble types v = (a, d, Q, c ), v = (a, d, Q, c ). v v, and B are the same. If the scenaro s that v s bdder s true type, accordng to the truthfulness, we have a [p (t, v ) c q (t, v )]dt d a [p (t, v ) c q (t, v )]dt If the scenaro s that v s bdder s true type, we have d a [p (t, v ) c q (t, v )]dt a [p (t, v ) c q (t, v )]dt By addng the two nequaltes above and usng c > c, we have d a q (t, v )dt > d q a (t, v )dt. Therefore q s monotone.

Fault tolerance in cloud technologies presented as a service

Fault tolerance in cloud technologies presented as a service Internatonal Scentfc Conference Computer Scence 2015 Pavel Dzhunev, PhD student Fault tolerance n cloud technologes presented as a servce INTRODUCTION Improvements n technques for vrtualzaton and performance

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

Power-of-Two Policies for Single- Warehouse Multi-Retailer Inventory Systems with Order Frequency Discounts

Power-of-Two Policies for Single- Warehouse Multi-Retailer Inventory Systems with Order Frequency Discounts Power-of-wo Polces for Sngle- Warehouse Mult-Retaler Inventory Systems wth Order Frequency Dscounts José A. Ventura Pennsylvana State Unversty (USA) Yale. Herer echnon Israel Insttute of echnology (Israel)

More information

Institute of Informatics, Faculty of Business and Management, Brno University of Technology,Czech Republic

Institute of Informatics, Faculty of Business and Management, Brno University of Technology,Czech Republic Lagrange Multplers as Quanttatve Indcators n Economcs Ivan Mezník Insttute of Informatcs, Faculty of Busness and Management, Brno Unversty of TechnologCzech Republc Abstract The quanttatve role of Lagrange

More information

Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs

Online Auctions in IaaS Clouds: Welfare and Profit Maximization with Server Costs Onlne Auctons n IaaS Clouds: Welfare and roft Maxmzaton wth Server Costs aox Zhang Dept. of Computer Scence The Unvety of Hong Kong xxzhang@cs.hku.hk Zongpeng L Dept. of Computer Scence Unvety of Calgary

More information

Answer: A). There is a flatter IS curve in the high MPC economy. Original LM LM after increase in M. IS curve for low MPC economy

Answer: A). There is a flatter IS curve in the high MPC economy. Original LM LM after increase in M. IS curve for low MPC economy 4.02 Quz Solutons Fall 2004 Multple-Choce Questons (30/00 ponts) Please, crcle the correct answer for each of the followng 0 multple-choce questons. For each queston, only one of the answers s correct.

More information

Feasibility of Using Discriminate Pricing Schemes for Energy Trading in Smart Grid

Feasibility of Using Discriminate Pricing Schemes for Energy Trading in Smart Grid Feasblty of Usng Dscrmnate Prcng Schemes for Energy Tradng n Smart Grd Wayes Tushar, Chau Yuen, Bo Cha, Davd B. Smth, and H. Vncent Poor Sngapore Unversty of Technology and Desgn, Sngapore 138682. Emal:

More information

Profit-Maximizing Virtual Machine Trading in a Federation of Selfish Clouds

Profit-Maximizing Virtual Machine Trading in a Federation of Selfish Clouds Proft-Maxmzng Vrtual Machne Tradng n a Federaton of Selfsh Clouds Hongxng L, Chuan Wu, Zongpeng L and Francs CM Lau Department of Computer Scence, The Unversty of Hong Kong, Hong Kong, Emal: hxl, cwu,

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

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

Economic-Robust Transmission Opportunity Auction in Multi-hop Wireless Networks

Economic-Robust Transmission Opportunity Auction in Multi-hop Wireless Networks Economc-Robust Transmsson Opportunty Aucton n Mult-hop Wreless Networks Mng L, Pan L, Mao Pan, and Jnyuan Sun Department of Electrcal and Computer Engneerng, Msssspp State Unversty, Msssspp State, MS 39762

More information

A Novel Auction Mechanism for Selling Time-Sensitive E-Services

A Novel Auction Mechanism for Selling Time-Sensitive E-Services A ovel Aucton Mechansm for Sellng Tme-Senstve E-Servces Juong-Sk Lee and Boleslaw K. Szymansk Optmaret Inc. and Department of Computer Scence Rensselaer Polytechnc Insttute 110 8 th Street, Troy, Y 12180,

More information

On Competitive Nonlinear Pricing

On Competitive Nonlinear Pricing On Compettve Nonlnear Prcng Andrea Attar Thomas Marott Franços Salané February 27, 2013 Abstract A buyer of a dvsble good faces several dentcal sellers. The buyer s preferences are her prvate nformaton,

More information

AN APPOINTMENT ORDER OUTPATIENT SCHEDULING SYSTEM THAT IMPROVES OUTPATIENT EXPERIENCE

AN APPOINTMENT ORDER OUTPATIENT SCHEDULING SYSTEM THAT IMPROVES OUTPATIENT EXPERIENCE AN APPOINTMENT ORDER OUTPATIENT SCHEDULING SYSTEM THAT IMPROVES OUTPATIENT EXPERIENCE Yu-L Huang Industral Engneerng Department New Mexco State Unversty Las Cruces, New Mexco 88003, U.S.A. Abstract Patent

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

Multi-Resource Fair Allocation in Heterogeneous Cloud Computing Systems

Multi-Resource Fair Allocation in Heterogeneous Cloud Computing Systems 1 Mult-Resource Far Allocaton n Heterogeneous Cloud Computng Systems We Wang, Student Member, IEEE, Ben Lang, Senor Member, IEEE, Baochun L, Senor Member, IEEE Abstract We study the mult-resource allocaton

More information

J. Parallel Distrib. Comput.

J. Parallel Distrib. Comput. J. Parallel Dstrb. Comput. 71 (2011) 62 76 Contents lsts avalable at ScenceDrect J. Parallel Dstrb. Comput. journal homepage: www.elsever.com/locate/jpdc Optmzng server placement n dstrbuted systems n

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

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

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 wangtngzhong2@sna.cn 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

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

When Network Effect Meets Congestion Effect: Leveraging Social Services for Wireless Services

When Network Effect Meets Congestion Effect: Leveraging Social Services for Wireless Services When Network Effect Meets Congeston Effect: Leveragng Socal Servces for Wreless Servces aowen Gong School of Electrcal, Computer and Energy Engeerng Arzona State Unversty Tempe, AZ 8587, USA xgong9@asuedu

More information

General Auction Mechanism for Search Advertising

General Auction Mechanism for Search Advertising General Aucton Mechansm for Search Advertsng Gagan Aggarwal S. Muthukrshnan Dávd Pál Martn Pál Keywords game theory, onlne auctons, stable matchngs ABSTRACT Internet search advertsng s often sold by an

More information

Demand Response of Data Centers: A Real-time Pricing Game between Utilities in Smart Grid

Demand Response of Data Centers: A Real-time Pricing Game between Utilities in Smart Grid Demand Response of Data Centers: A Real-tme Prcng Game between Utltes n Smart Grd Nguyen H. Tran, Shaole Ren, Zhu Han, Sung Man Jang, Seung Il Moon and Choong Seon Hong Department of Computer Engneerng,

More information

Economic Models for Cloud Service Markets

Economic Models for Cloud Service Markets Economc Models for Cloud Servce Markets Ranjan Pal and Pan Hu 2 Unversty of Southern Calforna, USA, rpal@usc.edu 2 Deutsch Telekom Laboratores, Berln, Germany, pan.hu@telekom.de Abstract. Cloud computng

More information

A Resource-trading Mechanism for Efficient Distribution of Large-volume Contents on Peer-to-Peer Networks

A Resource-trading Mechanism for Efficient Distribution of Large-volume Contents on Peer-to-Peer Networks A Resource-tradng Mechansm for Effcent Dstrbuton of Large-volume Contents on Peer-to-Peer Networks SmonG.M.Koo,C.S.GeorgeLee, Karthk Kannan School of Electrcal and Computer Engneerng Krannet School of

More information

Week 6 Market Failure due to Externalities

Week 6 Market Failure due to Externalities Week 6 Market Falure due to Externaltes 1. Externaltes n externalty exsts when the acton of one agent unavodably affects the welfare of another agent. The affected agent may be a consumer, gvng rse to

More information

1 Approximation Algorithms

1 Approximation Algorithms CME 305: Dscrete Mathematcs and Algorthms 1 Approxmaton Algorthms In lght of the apparent ntractablty of the problems we beleve not to le n P, t makes sense to pursue deas other than complete solutons

More information

Dynamic Online-Advertising Auctions as Stochastic Scheduling

Dynamic Online-Advertising Auctions as Stochastic Scheduling Dynamc Onlne-Advertsng Auctons as Stochastc Schedulng Isha Menache and Asuman Ozdaglar Massachusetts Insttute of Technology {sha,asuman}@mt.edu R. Srkant Unversty of Illnos at Urbana-Champagn rsrkant@llnos.edu

More information

A Lyapunov Optimization Approach to Repeated Stochastic Games

A Lyapunov Optimization Approach to Repeated Stochastic Games PROC. ALLERTON CONFERENCE ON COMMUNICATION, CONTROL, AND COMPUTING, OCT. 2013 1 A Lyapunov Optmzaton Approach to Repeated Stochastc Games Mchael J. Neely Unversty of Southern Calforna http://www-bcf.usc.edu/

More information

Dominant Resource Fairness in Cloud Computing Systems with Heterogeneous Servers

Dominant Resource Fairness in Cloud Computing Systems with Heterogeneous Servers 1 Domnant Resource Farness n Cloud Computng Systems wth Heterogeneous Servers We Wang, Baochun L, Ben Lang Department of Electrcal and Computer Engneerng Unversty of Toronto arxv:138.83v1 [cs.dc] 1 Aug

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

Online Mechanism Design for Electric Vehicle Charging

Online Mechanism Design for Electric Vehicle Charging Onlne Mechansm Desgn for Electrc Vehcle Chargng Enrco H. Gerdng eg@ecs.soton.ac.uk Davd C. Parkes parkes@eecs.harvard.edu Valentn Robu vr2@ecs.soton.ac.uk Alex Rogers acr@ecs.soton.ac.uk Unversty of Southampton,

More information

A hybrid global optimization algorithm based on parallel chaos optimization and outlook algorithm

A hybrid global optimization algorithm based on parallel chaos optimization and outlook algorithm Avalable onlne www.ocpr.com Journal of Chemcal and Pharmaceutcal Research, 2014, 6(7):1884-1889 Research Artcle ISSN : 0975-7384 CODEN(USA) : JCPRC5 A hybrd global optmzaton algorthm based on parallel

More information

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

Survey on Virtual Machine Placement Techniques in Cloud Computing Environment

Survey on Virtual Machine Placement Techniques in Cloud Computing Environment Survey on Vrtual Machne Placement Technques n Cloud Computng Envronment Rajeev Kumar Gupta and R. K. Paterya Department of Computer Scence & Engneerng, MANIT, Bhopal, Inda ABSTRACT In tradtonal data center

More information

Dynamic Resource Allocation and Power Management in Virtualized Data Centers

Dynamic Resource Allocation and Power Management in Virtualized Data Centers Dynamc Resource Allocaton and Power Management n Vrtualzed Data Centers Rahul Urgaonkar, Ulas C. Kozat, Ken Igarash, Mchael J. Neely urgaonka@usc.edu, {kozat, garash}@docomolabs-usa.com, mjneely@usc.edu

More information

Recurrence. 1 Definitions and main statements

Recurrence. 1 Definitions and main statements Recurrence 1 Defntons and man statements Let X n, n = 0, 1, 2,... be a MC wth the state space S = (1, 2,...), transton probabltes p j = P {X n+1 = j X n = }, and the transton matrx P = (p j ),j S def.

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

On the Interaction between Load Balancing and Speed Scaling

On the Interaction between Load Balancing and Speed Scaling On the Interacton between Load Balancng and Speed Scalng Ljun Chen and Na L Abstract Speed scalng has been wdely adopted n computer and communcaton systems, n partcular, to reduce energy consumpton. An

More information

Optimal Bidding Strategies for Generation Companies in a Day-Ahead Electricity Market with Risk Management Taken into Account

Optimal Bidding Strategies for Generation Companies in a Day-Ahead Electricity Market with Risk Management Taken into Account Amercan J. of Engneerng and Appled Scences (): 8-6, 009 ISSN 94-700 009 Scence Publcatons Optmal Bddng Strateges for Generaton Companes n a Day-Ahead Electrcty Market wth Rsk Management Taken nto Account

More information

A Secure Password-Authenticated Key Agreement Using Smart Cards

A Secure Password-Authenticated Key Agreement Using Smart Cards A Secure Password-Authentcated Key Agreement Usng Smart Cards Ka Chan 1, Wen-Chung Kuo 2 and Jn-Chou Cheng 3 1 Department of Computer and Informaton Scence, R.O.C. Mltary Academy, Kaohsung 83059, Tawan,

More information

Dynamic Pricing for Smart Grid with Reinforcement Learning

Dynamic Pricing for Smart Grid with Reinforcement Learning Dynamc Prcng for Smart Grd wth Renforcement Learnng Byung-Gook Km, Yu Zhang, Mhaela van der Schaar, and Jang-Won Lee Samsung Electroncs, Suwon, Korea Department of Electrcal Engneerng, UCLA, Los Angeles,

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

Joint Resource Allocation and Base-Station. Assignment for the Downlink in CDMA Networks

Joint Resource Allocation and Base-Station. Assignment for the Downlink in CDMA Networks Jont Resource Allocaton and Base-Staton 1 Assgnment for the Downlnk n CDMA Networks Jang Won Lee, Rav R. Mazumdar, and Ness B. Shroff School of Electrcal and Computer Engneerng Purdue Unversty West Lafayette,

More information

Price Competition in an Oligopoly Market with Multiple IaaS Cloud Providers

Price Competition in an Oligopoly Market with Multiple IaaS Cloud Providers Prce Competton n an Olgopoly Market wth Multple IaaS Cloud Provders Yuan Feng, Baochun L, Bo L Department of Computng, Hong Kong Polytechnc Unversty Department of Electrcal and Computer Engneerng, Unversty

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 reprnts@benthamscence.ae 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

On the Interaction between Load Balancing and Speed Scaling

On the Interaction between Load Balancing and Speed Scaling On the Interacton between Load Balancng and Speed Scalng Ljun Chen, Na L and Steven H. Low Engneerng & Appled Scence Dvson, Calforna Insttute of Technology, USA Abstract Speed scalng has been wdely adopted

More information

THE DISTRIBUTION OF LOAN PORTFOLIO VALUE * Oldrich Alfons Vasicek

THE DISTRIBUTION OF LOAN PORTFOLIO VALUE * Oldrich Alfons Vasicek HE DISRIBUION OF LOAN PORFOLIO VALUE * Oldrch Alfons Vascek he amount of captal necessary to support a portfolo of debt securtes depends on the probablty dstrbuton of the portfolo loss. Consder a portfolo

More information

Calculation of Sampling Weights

Calculation of Sampling Weights Perre Foy Statstcs Canada 4 Calculaton of Samplng Weghts 4.1 OVERVIEW The basc sample desgn used n TIMSS Populatons 1 and 2 was a two-stage stratfed cluster desgn. 1 The frst stage conssted of a sample

More information

Performance Analysis of Energy Consumption of Smartphone Running Mobile Hotspot Application

Performance Analysis of Energy Consumption of Smartphone Running Mobile Hotspot Application Internatonal Journal of mart Grd and lean Energy Performance Analyss of Energy onsumpton of martphone Runnng Moble Hotspot Applcaton Yun on hung a chool of Electronc Engneerng, oongsl Unversty, 511 angdo-dong,

More information

J. Parallel Distrib. Comput. Environment-conscious scheduling of HPC applications on distributed Cloud-oriented data centers

J. Parallel Distrib. Comput. Environment-conscious scheduling of HPC applications on distributed Cloud-oriented data centers J. Parallel Dstrb. Comput. 71 (2011) 732 749 Contents lsts avalable at ScenceDrect J. Parallel Dstrb. Comput. ournal homepage: www.elsever.com/locate/pdc Envronment-conscous schedulng of HPC applcatons

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

Schedulability Bound of Weighted Round Robin Schedulers for Hard Real-Time Systems

Schedulability Bound of Weighted Round Robin Schedulers for Hard Real-Time Systems Schedulablty Bound of Weghted Round Robn Schedulers for Hard Real-Tme Systems Janja Wu, Jyh-Charn Lu, and We Zhao Department of Computer Scence, Texas A&M Unversty {janjaw, lu, zhao}@cs.tamu.edu Abstract

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

1 Example 1: Axis-aligned rectangles

1 Example 1: Axis-aligned rectangles COS 511: Theoretcal Machne Learnng Lecturer: Rob Schapre Lecture # 6 Scrbe: Aaron Schld February 21, 2013 Last class, we dscussed an analogue for Occam s Razor for nfnte hypothess spaces that, n conjuncton

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

Optimal Customized Pricing in Competitive Settings

Optimal Customized Pricing in Competitive Settings Optmal Customzed Prcng n Compettve Settngs Vshal Agrawal Industral & Systems Engneerng, Georga Insttute of Technology, Atlanta, Georga 30332 vshalagrawal@gatech.edu Mark Ferguson College of Management,

More information

Modeling and Analysis of 2D Service Differentiation on e-commerce Servers

Modeling and Analysis of 2D Service Differentiation on e-commerce Servers Modelng and Analyss of D Servce Dfferentaton on e-commerce Servers Xaobo Zhou, Unversty of Colorado, Colorado Sprng, CO zbo@cs.uccs.edu Janbn We and Cheng-Zhong Xu Wayne State Unversty, Detrot, Mchgan

More information

Efficient Project Portfolio as a tool for Enterprise Risk Management

Efficient Project Portfolio as a tool for Enterprise Risk Management Effcent Proect Portfolo as a tool for Enterprse Rsk Management Valentn O. Nkonov Ural State Techncal Unversty Growth Traectory Consultng Company January 5, 27 Effcent Proect Portfolo as a tool for Enterprse

More information

An Analysis of Central Processor Scheduling in Multiprogrammed Computer Systems

An Analysis of Central Processor Scheduling in Multiprogrammed Computer Systems STAN-CS-73-355 I SU-SE-73-013 An Analyss of Central Processor Schedulng n Multprogrammed Computer Systems (Dgest Edton) by Thomas G. Prce October 1972 Techncal Report No. 57 Reproducton n whole or n part

More information

Marginal Revenue-Based Capacity Management Models and Benchmark 1

Marginal Revenue-Based Capacity Management Models and Benchmark 1 Margnal Revenue-Based Capacty Management Models and Benchmark 1 Qwen Wang 2 Guanghua School of Management, Pekng Unversty Sherry Xaoyun Sun 3 Ctgroup ABSTRACT To effcently meet customer requrements, a

More information

Economic Models for Cloud Service Markets Pricing and Capacity Planning

Economic Models for Cloud Service Markets Pricing and Capacity Planning Economc Models for Cloud Servce Markets Prcng and Capacty Plannng Ranjan Pal 1 and Pan Hu 2 1 Unversty of Southern Calforna, USA, rpal@usc.edu 2 Deutsch Telekom Laboratores, Berln, Germany, pan.hu@telekom.de

More information

Allocating Collaborative Profit in Less-than-Truckload Carrier Alliance

Allocating Collaborative Profit in Less-than-Truckload Carrier Alliance J. Servce Scence & Management, 2010, 3: 143-149 do:10.4236/jssm.2010.31018 Publshed Onlne March 2010 (http://www.scrp.org/journal/jssm) 143 Allocatng Collaboratve Proft n Less-than-Truckload Carrer Allance

More information

Efficient On-Demand Data Service Delivery to High-Speed Trains in Cellular/Infostation Integrated Networks

Efficient On-Demand Data Service Delivery to High-Speed Trains in Cellular/Infostation Integrated Networks IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. XX, NO. XX, MONTH 2XX 1 Effcent On-Demand Data Servce Delvery to Hgh-Speed Trans n Cellular/Infostaton Integrated Networks Hao Lang, Student Member,

More information

8.5 UNITARY AND HERMITIAN MATRICES. The conjugate transpose of a complex matrix A, denoted by A*, is given by

8.5 UNITARY AND HERMITIAN MATRICES. The conjugate transpose of a complex matrix A, denoted by A*, is given by 6 CHAPTER 8 COMPLEX VECTOR SPACES 5. Fnd the kernel of the lnear transformaton gven n Exercse 5. In Exercses 55 and 56, fnd the mage of v, for the ndcated composton, where and are gven by the followng

More information

A Game-Theoretic Approach for Minimizing Security Risks in the Internet-of-Things

A Game-Theoretic Approach for Minimizing Security Risks in the Internet-of-Things A Game-Theoretc Approach for Mnmzng Securty Rsks n the Internet-of-Thngs George Rontds, Emmanoul Panaouss, Aron Laszka, Tasos Daguklas, Pasquale Malacara, and Tansu Alpcan Hellenc Open Unversty, Greece

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

Online Advertisement, Optimization and Stochastic Networks

Online Advertisement, Optimization and Stochastic Networks Onlne Advertsement, Optmzaton and Stochastc Networks Bo (Rambo) Tan and R. Srkant Department of Electrcal and Computer Engneerng Unversty of Illnos at Urbana-Champagn Urbana, IL, USA 1 arxv:1009.0870v6

More information

Supply network formation as a biform game

Supply network formation as a biform game Supply network formaton as a bform game Jean-Claude Hennet*. Sona Mahjoub*,** * LSIS, CNRS-UMR 6168, Unversté Paul Cézanne, Faculté Sant Jérôme, Avenue Escadrlle Normande Némen, 13397 Marselle Cedex 20,

More information

Forecasting the Direction and Strength of Stock Market Movement

Forecasting the Direction and Strength of Stock Market Movement Forecastng the Drecton and Strength of Stock Market Movement Jngwe Chen Mng Chen Nan Ye cjngwe@stanford.edu mchen5@stanford.edu nanye@stanford.edu Abstract - Stock market s one of the most complcated systems

More information

Chapter 4 ECONOMIC DISPATCH AND UNIT COMMITMENT

Chapter 4 ECONOMIC DISPATCH AND UNIT COMMITMENT Chapter 4 ECOOMIC DISATCH AD UIT COMMITMET ITRODUCTIO A power system has several power plants. Each power plant has several generatng unts. At any pont of tme, the total load n the system s met by the

More information

Addendum to: Importing Skill-Biased Technology

Addendum to: Importing Skill-Biased Technology Addendum to: Importng Skll-Based Technology Arel Bursten UCLA and NBER Javer Cravno UCLA August 202 Jonathan Vogel Columba and NBER Abstract Ths Addendum derves the results dscussed n secton 3.3 of our

More information

2. SYSTEM MODEL. the SLA (unlike the only other related mechanism [15] we can compare it is never able to meet the SLA).

2. SYSTEM MODEL. the SLA (unlike the only other related mechanism [15] we can compare it is never able to meet the SLA). Managng Server Energy and Operatonal Costs n Hostng Centers Yyu Chen Dept. of IE Penn State Unversty Unversty Park, PA 16802 yzc107@psu.edu Anand Svasubramanam Dept. of CSE Penn State Unversty Unversty

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 peter.vortsch@ptv.de Peter Möhl, PTV AG,

More information

CloudMedia: When Cloud on Demand Meets Video on Demand

CloudMedia: When Cloud on Demand Meets Video on Demand CloudMeda: When Cloud on Demand Meets Vdeo on Demand Yu Wu, Chuan Wu, Bo L, Xuanja Qu, Francs C.M. Lau Department of Computer Scence, The Unversty of Hong Kong, Emal: {ywu,cwu,xjqu,fcmlau}@cs.hku.hk Department

More information

Support Vector Machines

Support Vector Machines Support Vector Machnes Max Wellng Department of Computer Scence Unversty of Toronto 10 Kng s College Road Toronto, M5S 3G5 Canada wellng@cs.toronto.edu Abstract Ths s a note to explan support vector machnes.

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

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 pchuan@khu.ac.kr Jae Hyeok Son Scence and Engneerng, KyungHee Unversty, Korea sonaehyeok@khu.ac.kr

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

MAPP. MERIS level 3 cloud and water vapour products. Issue: 1. Revision: 0. Date: 9.12.1998. Function Name Organisation Signature Date

MAPP. MERIS level 3 cloud and water vapour products. Issue: 1. Revision: 0. Date: 9.12.1998. Function Name Organisation Signature Date 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

How Sets of Coherent Probabilities May Serve as Models for Degrees of Incoherence

How Sets of Coherent Probabilities May Serve as Models for Degrees of Incoherence 1 st Internatonal Symposum on Imprecse Probabltes and Ther Applcatons, Ghent, Belgum, 29 June 2 July 1999 How Sets of Coherent Probabltes May Serve as Models for Degrees of Incoherence Mar J. Schervsh

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

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

Cost Minimization using Renewable Cooling and Thermal Energy Storage in CDNs

Cost Minimization using Renewable Cooling and Thermal Energy Storage in CDNs Cost Mnmzaton usng Renewable Coolng and Thermal Energy Storage n CDNs Stephen Lee College of Informaton and Computer Scences UMass, Amherst stephenlee@cs.umass.edu Rahul Urgaonkar IBM Research rurgaon@us.bm.com

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

Discriminatory versus Uniform-Price Electricity Auctions with Supply Function Equilibrium

Discriminatory versus Uniform-Price Electricity Auctions with Supply Function Equilibrium Dscrmnatory versus Unform-Prce Electrcty Auctons wth Supply Functon Equlbrum Talat S. Genc a December, 007 Abstract. A goal of ths paper s to compare results for dscrmnatory auctons to results for unform-prce

More information

Chapter 7: Answers to Questions and Problems

Chapter 7: Answers to Questions and Problems 19. Based on the nformaton contaned n Table 7-3 of the text, the food and apparel ndustres are most compettve and therefore probably represent the best match for the expertse of these managers. Chapter

More information

Using Series to Analyze Financial Situations: Present Value

Using Series to Analyze Financial Situations: Present Value 2.8 Usng Seres to Analyze Fnancal Stuatons: Present Value In the prevous secton, you learned how to calculate the amount, or future value, of an ordnary smple annuty. The amount s the sum of the accumulated

More information

"Research Note" APPLICATION OF CHARGE SIMULATION METHOD TO ELECTRIC FIELD CALCULATION IN THE POWER CABLES *

Research Note APPLICATION OF CHARGE SIMULATION METHOD TO ELECTRIC FIELD CALCULATION IN THE POWER CABLES * Iranan Journal of Scence & Technology, Transacton B, Engneerng, ol. 30, No. B6, 789-794 rnted n The Islamc Republc of Iran, 006 Shraz Unversty "Research Note" ALICATION OF CHARGE SIMULATION METHOD TO ELECTRIC

More information

Optimization Model of Reliable Data Storage in Cloud Environment Using Genetic Algorithm

Optimization Model of Reliable Data Storage in Cloud Environment Using Genetic Algorithm Internatonal Journal of Grd Dstrbuton Computng, pp.175-190 http://dx.do.org/10.14257/gdc.2014.7.6.14 Optmzaton odel of Relable Data Storage n Cloud Envronment Usng Genetc Algorthm Feng Lu 1,2,3, Hatao

More information

Profit-Aware DVFS Enabled Resource Management of IaaS Cloud

Profit-Aware DVFS Enabled Resource Management of IaaS Cloud IJCSI Internatonal Journal of Computer Scence Issues, Vol. 0, Issue, No, March 03 ISSN (Prnt): 694-084 ISSN (Onlne): 694-0784 www.ijcsi.org 37 Proft-Aware DVFS Enabled Resource Management of IaaS Cloud

More information

Risk-based Fatigue Estimate of Deep Water Risers -- Course Project for EM388F: Fracture Mechanics, Spring 2008

Risk-based Fatigue Estimate of Deep Water Risers -- Course Project for EM388F: Fracture Mechanics, Spring 2008 Rsk-based Fatgue Estmate of Deep Water Rsers -- Course Project for EM388F: Fracture Mechancs, Sprng 2008 Chen Sh Department of Cvl, Archtectural, and Envronmental Engneerng The Unversty of Texas at Austn

More information

Decentralized Inventory Sharing with Asymmetric Information. Xinghao Yan Hui Zhao 1

Decentralized Inventory Sharing with Asymmetric Information. Xinghao Yan Hui Zhao 1 Decentralzed Inventory Sharng wth Asymmetrc Informaton Xnghao Yan Hu Zhao 1 xyan@vey.uwo.ca zhaoh@purdue.edu Rchard Ivey School of Busness The Unversty of Western Ontaro Krannert School of Management Purdue

More information

PSYCHOLOGICAL RESEARCH (PYC 304-C) Lecture 12

PSYCHOLOGICAL RESEARCH (PYC 304-C) Lecture 12 14 The Ch-squared dstrbuton PSYCHOLOGICAL RESEARCH (PYC 304-C) Lecture 1 If a normal varable X, havng mean µ and varance σ, s standardsed, the new varable Z has a mean 0 and varance 1. When ths standardsed

More information

A generalized hierarchical fair service curve algorithm for high network utilization and link-sharing

A generalized hierarchical fair service curve algorithm for high network utilization and link-sharing Computer Networks 43 (2003) 669 694 www.elsever.com/locate/comnet A generalzed herarchcal far servce curve algorthm for hgh network utlzaton and lnk-sharng Khyun Pyun *, Junehwa Song, Heung-Kyu Lee Department

More information

Capacity Reservation for Time-Sensitive Service Providers: An Application in Seaport Management

Capacity Reservation for Time-Sensitive Service Providers: An Application in Seaport Management Capacty Reservaton for Tme-Senstve Servce Provders: An Applcaton n Seaport Management L. Jeff Hong Department of Industral Engneerng and Logstcs Management The Hong Kong Unversty of Scence and Technology

More information

Efficient Striping Techniques for Variable Bit Rate Continuous Media File Servers æ

Efficient Striping Techniques for Variable Bit Rate Continuous Media File Servers æ Effcent Strpng Technques for Varable Bt Rate Contnuous Meda Fle Servers æ Prashant J. Shenoy Harrck M. Vn Department of Computer Scence, Department of Computer Scences, Unversty of Massachusetts at Amherst

More information

Activity Scheduling for Cost-Time Investment Optimization in Project Management

Activity Scheduling for Cost-Time Investment Optimization in Project Management PROJECT MANAGEMENT 4 th Internatonal Conference on Industral Engneerng and Industral Management XIV Congreso de Ingenería de Organzacón Donosta- San Sebastán, September 8 th -10 th 010 Actvty Schedulng

More information