Mobile to Mobile Computational Offloading in Multi-hop Cooperative Networks

Size: px
Start display at page:

Download "Mobile to Mobile Computational Offloading in Multi-hop Cooperative Networks"

Transcription

1 Mobe to Mobe Computatona Offoadng n Mut-hop Cooperatve Networks Con Funa, Crstano Tappareo, Wend Henzeman Department of Eectrca and Computer Engneerng, Unversty of Rochester, Rochester, NY, USA Ema: {frstname.astname}@rochester.edu Abstract As the number of mobe devces that natvey support ad hoc communcaton protocos ncrease, arge ad hoc networks can be created not ony to factate communcaton among the mobe devces, but aso to assst devces that are executng computatonay ntensve appcatons. Pror work has deveoped computaton offoadng systems for mobe devces, but ths work has focused excusvey on offoadng to snge hop neghbors, due n part to the practca chaenges of settng up mut-hop networks usng exstng ad hoc communcaton protocos. However, mtng the offoadng of computaton to one-hop neghbors nherenty restrcts the number of devces that can partcpate n the dstrbuted computaton. In ths paper, we propose and evauate the performance of computatona offoadng wthn a mut-hop cooperatve network, where mobe devces are abe to share the computatona oad wth a other nodes n the network. Addtonay, we present an teratve task assgnment agorthm that can optmze the assgnment of computatona tasks to devces n such a mut-hop cooperatve network, takng nto account the communcaton overhead of the mut-hop network. Expermenta resuts, obtaned from an mpementaton on Androd devces, are ntegrated wth an anaytca mode that enabes the evauaton of system performance under a varety of condtons. These expermenta and anaytc resuts demonstrate the beneft of enabng computaton offoadng to a devces n a mut-hop cooperatve network. I. INTRODUCTION Recent advancements n VLSI and MEMs have aowed for ncreasngy compcated processors to become cheaper and more accessbe to a varety of appcatons. For nstance, these processors have become so accessbe that manufacturers are even offerng ruggedzed devces that use these processors 1, whe st hodng certfcaton for certan envronmenta condtons, such as ow pressure or extreme temperature. Wth devces that can wthstand such harsh condtons, t s not hard to magne that when encounterng hazardous stuatons, a frst responder woud beneft greaty from havng the abty to take n data from the fed, process t, and make an nformed decson based on the resut. For exampe, frst responders mght be equpped wth ruggedzed smart devces and depoyed to the ste of an earthquake to ade n reef efforts, or a mtary patro mght be abe to perform a Sgna Integence (SIGINT) appcaton for fndng mprovsed exposve devces (IEDs) 3. Typcay, these knds of appcatons requre mmense processng capabtes and Ths work was supported n part by Harrs Corporaton, RF Communcatons Dvson and n part by CEIS, an Empre State Deveopment desgnated Center for Advanced Technoogy. consume arge amounts of the devce s battery, f the devce s even abe to perform such computatons. One potenta souton s for the mobe devce to offoad these computatonay ntensve tasks to a remote server 4,, ether by settng up reay ponts throughout the area of nterest or usng satete communcaton 6. System ke MAUI 4 and ConeCoud are based on a remote executon scheme, where devces offoad computatonay ntensve portons of an appcaton to a remote server that returns the computed resut. Athough the aforementoned remote executon systems exhbt potenta for massve performance gans, oftentmes mtary or dsaster reef scenaros do not aow for the uxury of beng abe to permt hgh atences assocated wth offoadng computaton to remote servers, ether due to bandwdth or envronmenta constrants. Addtonay, n these scenaros, the devces may not be connected to a remote server at any gven tme. The concept of cyber foragng, where devces use nearby shared computng resources, was presented to provde an aternatve to the hgh atences assocated wth offoadng computaton to remote servers 7. Whe cyber foragng s presented n the context of pubc, untrustworthy and unmanaged computng resources, ths practce coud be transated to mtary or dsaster reef scenaros provded that the resources are depoyed over an area of nterest a-pror. On the other hand, Hyrax 8 proved the beneft of utzng a of the nodes n a network. By portng Hadoop Apache to mobe devces, the authors were abe to create a dstrbuted computng fe system and enabe dstrbuted computng on mobe devces. Hyrax acheves ths by utzng a centra server and nfrastructure based communcatons. Startng wth the work presented n 8, severa works presented soutons that are abe to offoad computaton to other nearby mobe devces n a compete ad hoc manner Honeybee, for exampe, provdes a programmng framework to ade mobe to mobe computatona offoadng, usng Buetooth to factate the communcaton between devces 9. More recenty, the authors n 1, 11 proposed and mpemented a mobe computatona offoadng system usng Androd s mpementaton of WF Drect. Whe these works have shown the feasbty of offoadng to mobe devces n an ad hoc settng, ther mpementatons are mted to a snge hop ad hoc network. The next step n the evouton of mobe computaton offoadng s to consder a the computatona resources avaabe n a mut-hop ad hoc

2 network. In ths regard, Serendpty enabes remote computng among mobe devces over ntermttenty connected ad hoc networks 1, usng a technque smar to those used n dsrupton toerant networks 13. By utzng a modfed verson of Djkstra s routng agorthm, the authors presented a water fng based task assgnment scheme, that uses an estmate of the tme requred to transfer data between two nodes to cacuate the number of tasks to assgn to a partcuar node. Serendpty s performance has aso been verfed through an mpementaton on Androd devces. Gven the above, our work ams at overcomng the mtatons of the prevous work by expotng a the computatona resources avaabe n a mut-hop ad hoc network. In partcuar, our contrbutons n ths paper are: Usng an mpementaton on Androd devces, we demonstrate the beneft of expandng the number of computatona resources that can be utzed for task competon, and show that n a rea appcaton, a smpe task dstrbuton s abe to reduce the tota tme of the dstrbuted computaton. We deveop a provaby optma agorthm for dstrbutng tasks to nodes wthn a mut-hop network that can be used to estmate the performance of offoadng the computaton over a mut-hop ad hoc network under dfferent settngs. Hence, the overa concuson of ths paper s that not ony s t feasbe to offoad computaton to nodes that are mutpe hops away, but there s aso a measurabe performance gan that can be acheved by dong so. The rest of the paper s organzed as foows. In Secton II, we descrbe our mobe to mobe computatona offoadng system, and motvate the need for an ntegent task assgnment. In Secton III, we present an anaytca mode for cacuatng the amount of tme and energy that s requred to assgn tasks to dfferent nodes n a mut-hop network. In Secton IV, we defne an objectve functon representng the cost to compute some number of tasks, as we as a provaby optma teratve agorthm for optmzng ths objectve functon. Secton V presents measured and anaytc resuts for computaton offoadng n a mut-hop WF Drect network. Secton VI concudes the paper. II. MOTIVATION Our mobe to mobe computatona offoadng system s composed of a set of N mobe devces, organzed nto a network through ad hoc communcatons. The devces are cooperatve, meanng that we do not consder the presence of any sefsh user. Moreover, we assume the exstence of a goba routng protoco, and each node has knowedge of a other avaabe nodes wthn the network. At a certan pont n tme, one of the devces needs to run a compex appcaton, whch can be dvded nto ceary defned homogenous tasks, as s the case for a mtary SIGINT appcaton 3, or non-nvasvey determnng the structura ntegrty of a budng durng dsaster reef efforts 14. These tasks are ndependent from each other, and ther computaton Gan over offoadng to one neghbor % Group -Hops 3-Hops Group -Hops Network 3-Hops Network Unform Group Unform -Hops Network Unform 3-Hops Network Fg. 1. Expermenta measurement of the percent gan over offoadng to ony a snge neghbor of a smpe greedy and unform task dstrbuton scheme. can be paraezed by offoadng the tasks to dfferent devces. The resuts of a the tasks executon s then merged n order to compete the nta compex appcaton. As a resut, assgnng a task to a devce requres the transmsson and recepton of the task and the resuts of the task executon over the ad hoc route. Ths requres communcaton energy consumpton on a the nodes aong the routng path, and ncurs a tme deay due to the propagaton of the data throughout the network. In order to determne the beneft of utzng mut-hop mobe to mobe computaton offoadng, we mpemented a computatona offoadng system on Androd devces, utzng the WF Drect protoco 1. We note that the functonates for creatng mut-group networks usng WF Drect are not natvey mpemented n Androd 16. Nevertheess, the work presented n 16, 17 provde dfferent soutons (e.g., tme sharng between groups, broadcastng and mutcastng data between groups and modfcatons of the Androd OS) to seamessy enabe mut-group communcaton usng stock and non-stock Androd devces. Gven the above, we deveoped a testbed by nterconnectng severa 13 Nexus 7s (N7), usng the modfed verson of Androd presented n 16. The 13 N7 has 16 GB of storage, GB of memory, and a 1.GHz quad-core Snapdragon CPU 18. For a the resuts presented n ths paper, we mted each devce to compute ony one task at a tme, and the devce coud not request the next task unt t returned the resut. Each of the presented data ponts are an average of thrty tras. A of the measurements were taken n an ndoor workspace, and the nodes were statonary durng these experments. We acknowedge that n rea scenaros, nodes are generay mobe, and may enter or eave a network. However, to accuratey determne the mpact of the task assgnment and to mt the varabty across dfferent experments, we kept the nodes statonary. Moreover, a the resuts of ths paper have been obtaned wth the screen turned off. To evauate the performance of the envsoned mut-hop mobe to mobe computaton offoadng system, n Fgure 1 we compare the expermenta performance gan to compete a set of tasks reatve to a task dstrbuton that ony utzes one addtona devce wth extendng the computaton to a arger network (see Fgure 1). For assgnng the tasks, we mpemented both a smpe task dstrbuton, where

3 each devce n the network requests a new task as soon as t has competed the prevous task, and a Unform task dstrbuton, where the tasks are unformy dstrbuted throughout the network. Whe both task assgnments provde substanta performance gan compared to offoadng to a snge neghbor when the computaton tme s much arger then the tme requred to assgn the tasks and receve the resuts, extendng the computaton to the mut-hop network can be detrmenta when the communcaton tme s greater than the computaton. As a resut, t s not straghtforward to determne how to best assgn the tasks due to the communcaton costs of dstrbutng the tasks. Moreover, whe both the Unform and the smpe task dstrbuton aready provde gans n some cases, t s not possbe to drecty determne the performance of a dstrbuton before assgnng the tasks and, at the same tme, t s not cear f further performance mprovements are even possbe. Thus, n the next sectons we frst mode the dfferent eements of the system under consderaton and then present an teratve agorthm that s guaranteed to return the optma task assgnment. III. SYSTEM MODEL In ths secton we present the detas of the dfferent components of the system descrbed n Secton II, whch are then used to derve the optma task dstrbuton pocy presented n Secton IV. A. Task Tme Mode We start by consderng the tota tme D requred to compute a gven task at a partcuar devce. We defne ths tota tme as the tme between the task assgnment and the end of the transmsson of the resut of the task executon. Thus, n order to determne the tme requred to assgn, compute and then receve the resuts of the task executon we need to consder two man components, namey the communcaton tme requred to dstrbute the task and receve the resuts, and the tme actuay spent for the task executon. In order to derve the communcaton tme, we consder that each task s characterzed by three parameters, namey the task data sze t, the reatve resut data sze r and the task compexty c. Thus, accordng to ths defnton, assgnng a task(t, c, r) to a devce requres the transmsson/recepton of t bts, whe the resut of the task executon to be reported back to the task generator entas the transmsson/recepton of r bts. The tme requred to execute a snge task depends on the CPU of the mobe devce and on the compexty c of the task to be executed, whch, wthout oss of generaty, can be consdered as a functon e (c). Therefore, the tota tme requred to compute a snge task at devce depends on the devce s computatona capabtes, and on the characterstcs of the rado technoogy used for the data exchange, as we as on the number of hops H between the devce that generated the tasks and devce. Gven the above, the tota deay D (t,c,t) experenced by a task (t,c,r) assgned to devce s gven by ( H D (t,c,r) = t + r ) H 1 + T sw +e (c), (1) wheret tx represents the transmsson throughput of the communcaton hop, represents the communcaton hop n the reverse path, and T sw accounts for addtona swtchng tme when routng the data between two subsequent communcaton nks as can be the case, for exampe, wth the WF Drect protoco (see, e.g., 16). We note that the transmsson throughput aso accounts for parae use of the same communcaton nk (by, e.g., the smutaneous assgnment of dfferent tasks) and, snce the throughput on wreess networks fuctuates due to the sgna strength and channe mparments, assgnng vaues to the transmsson throughput ony serves as an exampe of the achevabe performance of a network. B. Task Energy Mode In most cases, the system descrbed n ths paper w consst of battery operated mobe devces. As a resut, t s mportant not ony to characterze the deay experenced by the computaton of each task, but aso to determne the mpact n terms of energy consumpton of the task dstrbuton. Therefore, n what foows we defne the tota energy consumpton requred to assgn a task to a partcuar devce. Smar to the task tme mode defned n the prevous secton, dstrbutng a task entas a communcaton energy consumpton and the task executon energy consumpton. In order to defne the communcaton energy mode, we frst defnep tx andp rx to be the transmsson and recepton power consumpton of a partcuar ad hoc communcaton nk, respectvey. We note that a partcuar nk actuay represents a par of nodes (h, k), where h s the devce transmttng the data and k s the devce recevng the data. Thus, wth a tte abuse of notaton we can wrte P tx = Ph tx and P rx = Pk rx, where Ptx h represents the power consumpton of mobe devce h durng transmsson, and Pk rx represents the power consumpton of mobe devce k durng recepton. In the same way, we defne P ex to represent the power consumpton of mobe devce durng task computaton. As a resut, we defne the tota energy expendture when assgnng a task (t,c,r) to devce as H E (t,c,r) = t H 1 + (P tx +P rx )+ r E sw +P ex e (c), (P tx +P rx ) where, smar to Eq. (1),represents the communcaton hop n the reverse path (e.g., f represents the par of nodes(h,k), refers to the par (k,h)), and E sw represents the amount of energy requred for swtchng when routng the data between two subsequent communcaton nks. IV. OPTIMAL TASK DISTRIBUTION The goa of our anayss s to fnd the optma task dstrbuton pocy that mnmzes an overa cost metrc by utzng a the avaabe nodes n a mut-hop network. In the foowng sectons, we frst formuate the optma tasks dstrbuton ()

4 probem, and then we present an teratve agorthm that s abe to fnd the optma task assgnment. A. Optmzaton Probem Let N be the number of mobe devces partcpatng n the dstrbuted computaton and M be the number of homogeneous tasks 1 to be dstrbuted. We defne φ to be a vector such that φ = φ,φ 1,...,φ N, where φ represents the number of tasks assgned to a partcuar devce, such that φ Z +, φ M and N =1 φ = M, and Φ to be the set of a the possbe task assgnments that satsfy these condtons. Moreover, et C = C 1,C,...,C N be a cost vector, where C wth = 1,...,N represents the cost of assgnng a task to devce. Gven the above, our objectve s to sove the foowng optmzaton probem: mnf(φ), (3) φ Φ where F(φ) = max {φ C } represents the cost of a partcuar task assgnment φ Φ. We note that F(φ) accounts for the assgnment of tasks to botteneck nodes and s sutabe for, e.g, fndng the mnmum tme requred to perform the dstrbuted computaton of the M tasks. B. Proposed Pocy The optmzaton probem defned n Eq. (3) can be seen as a varaton of the Lnear Botteneck Assgnment Probem (LBAP) 19. Dfferent from a tradtona LBAP, we consder the possbty of assgnng to a partcuar agent (.e., devce) more than one task. We note that whe dfferent agorthms for sovng the LBAP probem have been proposed n the terature, e.g.,, aowng an agent to execute more than one task adds addtona compexty to the probem. At the same tme, the assumpton of homogeneous tasks aows us to smpfy the probem, whch can be optmay soved wth an teratve agorthm. In what foows, we frst propose an agorthm to sove probem (3), and then prove the optmaty of the souton determned by our agorthm. In partcuar, we am to fnd an optma pocy φ for probem (3). To ths end, et X α be a vector of ength N, so that X α represents the number of tasks that have been assgned to devce up to the assgnment of task α, wth α M. In addton, et Y be a decson vector of ength N, so that a but one of the eements of Y are zero. Gven the above, the agorthm for fndng the optma task assgnment s defned n Agorthm 1. C. Performance Anayss To prove that Agorthm 1 s abe to determne an optma souton to the optmzaton probem defned n Eq. (3), we frst show that startng from an optma task assgnment for K tasks, t s possbe to construct an optma task assgnment for K +1 tasks. 1 The extenson of the resuts of ths paper to the case of non-homogeneous tasks s consdered as future work. Agorthm 1 Cost Optma Task Dstrbuton 1: X ={,,,...,} : for Each Task α 1,...,M do 3: Y={,,,...,} 4: = argmn((x α 1 + 1) C) : Y = 1 6: X α = X α 1 +Y 7: end for Theorem IV.1. Gven an optma K tasks assgnment Γ K, an optma K +1 tasks assgnment Γ K+1 can be obtaned as Γ K+1 = argmn ψ Ψ K+1 max {ψ C = 1,...,N}, where Ψ K+1 = {ψ }, ψ = Γ K + ǫ and ǫ s an a-zeros vector wth a one n poston, for each = 1,...,N. Proof. We w prove ths n two steps. In the frst step, we consder the case where Γ K s the ony optma task assgnment for K tasks, whe n the second step, we consder the case of mutpe K task assgnments that have the same optma cost. For the case where there s a unque optma souton to dstrbute K tasks, et s assume by contradcton that Γ K+1 determned accordng to the above defnton s not optma. Ths, n turn, means that there exsts a K+1 tasks assgnment Θ K+1, such that Θ K+1 / Ψ K+1 and F(Θ K+1 ) < F(Γ K+1 ). (4) Moreover, we can consder Θ K+1 to be derved from a K tasks assgnment Θ K, such that F(Γ K ) < F(Θ K ) and F(Θ K ) F(Θ K+1 ). () As a resut, by combnng (4) and (), we obtan F(Θ K ) < F(Γ K+1 ). (6) On the other end, assumng that F(Θ K ) = Θ K m C m (.e., devce m s the botteneck devce that determnes the overa cost of the non-optma task assgnment Θ K ), we have that Γ K m < ΘK m. Thus, ΓK m +1 ΘK m. Now, f ΓK m +1 < ΘK m, then F(Γ K + e m ) = F(Γ K ) < F(Θ K+1 ), whch contradcts the hypothess n (4). If Γ K m +1 = ΘK m, nstead, F(ΓK +e m ) = F(Θ K ) F(Θ K+1 ), whch st contradcts the hypothess n (4). As a resut, f Γ K s the unque optma K tasks assgnment, then the K + 1 task assgnment Γ K+1 obtaned by Theorem IV.1 s an optma K +1 tasks assgnment. For the case where F(Γ K ) = F(Θ K ), there exst at east one Γ K m, m = 1,...,N such that Γ K m < Θ K m snce Γ K Θ K. Thus, Γ K m +1 ΘK m whch, foowng an argument smar to the one descrbe above, resuts n F(Γ K+1 ) = F(Γ K +e m ) F(Θ K+1 ), whch contradcts the hypothess n (4). Therefore, Γ K+1 obtaned by Theorem IV.1 s an optma K +1 tasks assgnment.

5 We w now use the resut of Theorem IV.1 to show that the agorthm presented n Secton IV-B returns an optma souton to the optmzaton probem (3). Theorem IV.. Agorthm 1 yeds an optma souton to the optmzaton probem defned n Eq. (3). Proof. In what foows, we w prove by nducton on the number of tasks M that the souton determned by Agorthm 1 s an optma souton to the optmzaton probem defned n Eq. (3). For the case M = 1, the set of a possbe tasks assgnment s represented by Φ = {1,,...,,,1,...,,...,,,...,1}; as a resut, we can easy see that C 1 opt = mn φǫφ max {φ C } = mn( 1 C), (7) where 1 s a vector of a ones. Thus, C 1 opt can further be smpfed to C 1 opt = C, (8) where = argmn( 1 C). Now to prove that our agorthm s capabe of yedng an optma souton for M = 1 task, et C ago be the cost of the souton produced by the agorthm, defned as C 1 ago = max { X 1 C }. (9) Snce X s ntazed to a vector of zeros, as shown n step of our agorthm, Cago 1 smpfes to Cago 1 = max {(X +1)C } {(Xj +)C j j }, (1) where X s the devce whch s chosen at steps 4 and of our agorthm, and X j s the subset of X that were not chosen at step 4 (.e., by the argmn functon). Eq. (1) can further be rewrtten as Cago 1 = max {1 C } { C j j } = C, (11) whch, compared wth Eq. (8), shows that Cago 1 =C1 opt. Let s now assume thatcago K = CK opt up tom = K, and that X K = Γ K, where Γ K s the optma souton for M = K. For the case where tasks M = K +1 we can wrte opt = mn φǫφ max {φ C }. (1) However, as shown n Theorem IV.1, we can restrct the search of the optma souton to the set Ψ K+1, whch has reduced cardnaty to N. Ths means that we can rewrte Eq. (1) as opt = mn ψǫψ K+1 max {ψ C }, (13) whch n turn can be expanded and utmatey yeds three possbe outcomes,.e., opt = mn Γ K g Cg S.T.(ΓK +1)C =Γ K g Cg (Γ K g +1)Cg (Γ h +1)C h hs.t.(γ K h +1)C h>γ K g Cg, (14) where Γ K g C g = C K opt, whch can be rewrtten as opt = max {(Γ K +1)C } {(Γ K j +)C j j }, (1) where (Γ K +1)C, = argmn((γ K +1)C). Now for the souton determned by the agorthm, we can wrte: ago = max{xk+1 C = 1,...,N}, (16) whch n turn can be smpfed to ago = max {(X K +1)C } {(X K j +)C j j } (17) whch, combned wth Eq. (1), returns ago = Copt K+1. Thus, accordng to Theorem IV.1, the K+1 tasks assgnment determned by the agorthm s guaranteed to be optma, whch concudes the proof. V. NUMERICAL RESULTS Usng the testbed descrbed n Secton II, we compared the task dstrbuton found usng the agorthm presented n Secton IV-B wth both the greedy and unform task dstrbuton agorthms descrbed prevousy. Each tra was started after the correct task dstrbuton, f appcabe, was found and was stopped when the ast resut of the computaton was receved. We note that, when optmzng for energy, the optmzaton probem presented n Secton IV can be smpfed to mnmze the tota energy consumpton of the task assgnment. Nevertheess, the cost-optma task dstrbuton presented n Secton IV mnmzes the maxmum energy consumpton requred for assgnng tasks to each devce, thus fary dstrbutng the tasks throughout the network. A. Performance Evauaton For a the resuts presented n ths secton, we consder a homogeneous network and defne the cost vector C = C 1,C,...,C N of the optmzaton probem n Eq. (3) such that C = D (t,c,r), where t = 13B, r {1KB,1KB,1MB} and c s determned so that e (c) {1ms,1ms,1s,1s}. Whe we do not ncude the energy consumpton at each devce (.e.,e n Eq. ()) n the defnton of the cost vector C, when our agorthm fnds mutpe task assgnments that resut n the same deay cost, we seect the task assgnment that mnmzes the overa network energy consumpton. Fnay, we note that dfferent defntons of the cost vector C are aso possbe. As an exampe, the cost C defned above can addtonay ncude a eve of reabty R,1, thus becomng C = D (t,c,r)/r. 1) Offoadng Resuts: In order to vadate our mode, n Fgures -4 we pot the tme requred to compute a set of tasks as a functon of dfferent computaton/communcaton tme ratos, for both an Androd mpementaton of the task dstrbuton as we as a smuaton of the task dstrbuton usng the mode descrbed n Secton III. In these fgures, the dashed nes represent the smuated performance of the task dstrbuton obtaned through the teratve agorthm descrbed,

6 Tme ms Iteratve Average Throughput Unform Average Throughput Iteratve Hgh Throughput Unform Hgh Throughput Iteratve Low Throughput Unform Low Throughput Impemented Iteratve Impemented Unform Impemented Expermenta Resuts Tme ms Iteratve Average Throughput Unform Average Throughput Iteratve Hgh Throughput Unform Hgh Throughput Iteratve Low Throughput Unform Low Throughput Impemented Iteratve Impemented Unform Impemented Expermenta Resuts e (c)=1s 1 e (c)=1ms e (c)=1ms e (c)=1s e (c)=1s 4 e (c)=1ms e (c)=1ms e (c)=1s Fg.. Impementaton and Smuaton resuts for dfferent computaton/communcaton ratos. Resut sze s 1 MB. Fg. 3. Impementaton and Smuaton resuts for dfferent computaton/communcaton ratos. Resut sze s 1 KB. n Secton IV-B, whe the contnuous nes represent the smuated performance of a unform task dstrbuton for dfferent vaues of transmsson throughputs. In partcuar, we set = {7Mbps,Tr avg,.mbps} n Eq. (1), wheretr avg s the average expermenta throughput measured for the dfferent data sze r (.e., T avg 1KB =.87Mbps, Tavg 1KB = 8.4Mbps and T avg 1MB = 3Mbps). The squares, damonds and stars, nstead, represent the resuts obtaned by mpementng the Iteratve, Unform and the task dstrbuton poces n our rea network of Androd devces, respectvey. The resuts n Fgures -4 show that when the computaton takes about tmes onger than the communcaton, a unform task dstrbuton provdes the same performance as the optma task dstrbuton. Beow ths pont, there are cear benefts n usng the task dstrbuton found wth our teratve agorthm, wth gans that become more evdent when the tme spent computng s comparabe to the communcaton tme. When the communcaton takes sgnfcanty onger than the computaton, the optma task dstrbuton can compete the set of tasks about twce as qucky as the unform task dstrbuton (.e., resut sze of 1MB and computaton tme of 1ms). Moreover, these resuts vadate the mode presented n Secton III, and show that the smuated resuts can provde a good approxmaton of the mpementaton resuts so ong as the approprate average nk throughputs are used. Hence, our teratve agorthm can be used to expore the performance of computaton offoadng n mut-hop ad hoc networks. The smpe task dstrbuton descrbed n Secton II s abe to adapt to the nstantaneous varatons n computatona tme (due to, e.g., operatng system operaton unreated to the actua task executon), as we as to the channe mparments that can severey affect the actua throughput of the communcaton nks. As a resut, the task dstrbuton s abe to attan performance very cose to the smuated performance of the optma task assgnment wth the hgh WF Drect throughput. When mnmzng the tota competon tme, thanks to ts adaptabty to the nstantaneous system varatons, the task dstrbuton outperforms the mpementaton of the optma task assgnment, thus makng t the de-facto choce for rea devce mpementatons. To gan further nsght nto the dfferences between the Tme ms Iteratve Average Throughput Unform Average Throughput Iteratve Hgh Throughput Unform Hgh Throughput Iteratve Low Throughput Unform Low Throughput Impemented Iteratve Impemented Unform Impemented e (c)=1ms Expermenta Resuts e (c)=1ms e (c)=1s e (c)=1s Fg. 4. Impementaton and Smuaton resuts for dfferent computaton/communcaton ratos. Resut sze s 1 KB. task dstrbuton and our teratve agorthm, n Fgure we compare the average task assgnments that were used n each case. As can be seen n Fgure (a), both the teratve and task dstrbuton schemes start assgnng tasks n a unform way when the computaton tme (1s) s much onger than the tme spent communcatng. When the stuaton s reversed (.e., communcaton s much onger than the 1ms computaton), nstead, n some cases no tasks are actuay assgned to the furthest node (see Fgure (d)). Overa, these resuts show that the throughput used to compute the optma task assgnment s over estmated, whch s partcuary evdent by the fact that the devce that s generatng the tasks (.e, the devce at hops), s most of the tme computng fewer tasks than when usng the approach. ) Benefts of Offoadng to Mut-hop Neghbors: The resuts presented n the prevous secton show that a task assgnment can adapt to changng computaton and communcaton envronments and hence can return a the tasks n the east amount of tme. However, our teratve agorthm provdes a task dstrbuton that s cose to that provded by the agorthm. Addtonay, the resuts n the prevous secton show that the smuated resuts match the mpementaton resuts and, as a consequence, smuatons based on the mode from Secton III can be used to provde nsght nto the performance of the system. Thus, n what foows we use the anaytca mode and the teratve agorthm descrbed n Sectons III-IV to further expore the gans that can be acheved by extendng the dstrbuted computaton to

7 hops 1 hop hops 3 hops 4 hops (a) 1s cacuaton Iteratve hops 1 hop hops 3 hops 4 hops (c) 1ms cacuaton Iteratve Iteratve hops 1 hop hops 3 hops 4 hops (b) 1s cacuaton Iteratve hops 1 hop hops 3 hops 4 hops (d) 1ms cacuaton Fg.. Task dstrbuton for the and Iteratve approach wth resut sze fxed to 1 MB. Gan over offoadng to one neghbor % Group -Hops 3-Hops 4-Hops Snge Group -Hops Network 3-Hops Network 4-Hops Network Fg. 6. Smuated measurement ndcatng the percent gan over offoadng to ony a snge neghbor. a the avaabe network resources. In partcuar, n order to hghght the benefts of offoadng the computaton to a the mobe devces n a network, n Fgure 6 we compare the gan n tme to compete a tasks reatve to a task dstrbuton that ony utzes one addtona devce (.e., snge-hop task dstrbuton as consdered n the terature 9 11) wth: a) offoadng the computaton to a of the nearest neghbors of the devce that s generatng the tasks (.e., Snge Group), and b) offoadng the computaton to a mut-hop network (.e., -hops, 3-hops and 4-hops network). As shown n Fgure 6, n ths smuaton we consder a treeke network wth the root node generatng the tasks havng four chdren nodes, and each subsequent chd servcng one addtona node. Ths network extends for, 3 or 4 hops, resutng n a tota of 4 devces n the source node s group that can be used for group task dstrbuton, and 8, 1 and 16 devces that can be used for network task dstrbuton. Fnay, Fgure 6 ceary shows the beneft of offoadng to a mut-hop network. In partcuar, offoadng to the arger network provdes up to 3% faster computaton tme than offoadng the tasks ony to the frst group, and a gan of up to 88% aganst offoadng to ony a snge devce (as s currenty supported n the terature). VI. CONCLUSIONS In ths paper we expored the benefts of enhancng mobe to mobe computatona offoadng n mut-hop cooperatve networks. By mpementng a mut-hop computatona offoadng system, we were abe to mpement dfferent task dstrbuton agorthms and verfy the accuracy of an anaytc mode and mpemented dfferent task dstrbuton strateges. Usng ths mode, we were abe to show the overa beneft of enabng offoadng to mut-hop neghbors n a network, whch can be qute arge when the tme to compute s much hgher than the tme to communcate the data. In our future work, we w expore the nter-reaton between reay nodes and nodes that are assgned tasks. In ths way, we can avod routng data through computaton bottenecks, as we mtng the number of tasks assgned to nodes that are pertnent for factatng communcaton. REFERENCES 1 Harrs RF-39-RT. Onne. Avaabe: capabtes/c4sr/ruggedzedtabet.asp MIL-STD-81G. Onne. Avaabe: MIL-STD-8-899/MIL-STD-81G M. T. Fynn, M. F. Pottnger, and P. D. Batcheor, Fxng nte: A bueprnt for makng ntegence reevant n afghanstan, DTIC Document, Tech. Rep., 1. 4 E. Cuervo, A. Baasubramanan, D.-k. Cho, A. Woman, S. Sarou, R. Chandra, and P. Bah, MAUI: Makng smartphones ast onger wth code offoad, n Proc. of ACM MobSys, 1, pp B.-G. Chun, S. Ihm, P. Manats, M. Nak, and A. Patt, ConeCoud: Eastc executon between mobe devce and coud, n Proc. of ACM EuroSys, 11, pp A. Khan, M. Othman, S. Madan, and S. Khan, A survey of mobe coud computng appcaton modes, Communcatons Surveys Tutoras, IEEE, vo. 16, no. 1, pp , Frst R. K. Baan, D. Gerge, M. Satyanarayanan, and J. Herbseb, Smpfyng cyber foragng for mobe devces, n Proc. of ACM MobSys, 7, pp E. E. Marne, Hyrax: Coud computng on mobe devces usng mapreduce, Sept N. Fernando, S. Loke, and W. Rahayu, Mobe crowd computng wth work steang, n Proc. of NBS, Mebourne, Austraa, Sept T. Penner, A. Johnson, B. Van Syke, M. Gurgus, and Q. Gu, Transent couds: Assgnment and coaboratve executon of tasks on mobe devces, n Prof. of IEEE GLOBECOM, Dec. 14, pp C. Funa, C. Tappareo, H. Ba, B. Karaogu, and W. Henzeman, Extendng vounteer computng through mobe ad hoc networkng, n Prof. of IEEE GLOBECOM, Dec. 14, pp C. Sh, V. Lakafoss, M. H. Ammar, and E. W. Zegura, Serendpty: Enabng remote computng among ntermttenty connected mobe devces, n Proc. of ACM MobHoc, 1, pp J. Burgess, B. Gaagher, D. Jensen, and B. N. Levne, MaxProp: Routng for vehce-based dsrupton-toerant networks. n Proc. of IEEE INFOCOM, Apr. 6, pp B. Raj, T. Jayakumar, and B. Rao, Non-destructve testng and evauaton for structura ntegrty, Sadhana, vo., no. 1, pp. 38, W-F Aance, PP Task Group, W-F Peer-to-Peer (PP) Technca Specfcaton, Verson 1., Dec C. Funa, C. Tappareo, and W. B. Henzeman, Supportng mut-hop devce-to-devce networks through WF drect mut-group networkng, ArXv e-prnts, Dec C. Casett, C. F. Chassern, L. Curto Pee, C. De Vae, Y. Duan, and P. Gaccone, Content-centrc routng n W-F Drect mut-group networks, ArXv e-prnts, Dec Asus Nexus 7 (13). Onne. Avaabe: Tabets Mobe/Nexus 7 13/ 19 D.-M. U. Dergs and U. Zmmermann, An augmentng path method for sovng near botteneck assgnment probems, Computng, vo. 19, no. 4, pp. 8 9, R. E. Burkard and U. Zmmermann, Weaky admssbe transformatons for sovng agebrac assgnment and transportaton probems, n Combnatora Optmzaton. Sprnger, 198, pp

MARKOV CHAIN AND HIDDEN MARKOV MODEL

MARKOV CHAIN AND HIDDEN MARKOV MODEL MARKOV CHAIN AND HIDDEN MARKOV MODEL JIAN ZHANG JIANZHAN@STAT.PURDUE.EDU Markov chan and hdden Markov mode are probaby the smpest modes whch can be used to mode sequenta data,.e. data sampes whch are not

More information

Achieving Optimal Throughput Utility and Low Delay with CSMA-like Algorithms: A Virtual Multi-Channel Approach

Achieving Optimal Throughput Utility and Low Delay with CSMA-like Algorithms: A Virtual Multi-Channel Approach Achevng Optma Throughput Utty and Low Deay wth SMA-ke Agorthms: A Vrtua Mut-hanne Approach Po-Ka Huang, Student Member, IEEE, and Xaojun Ln, Senor Member, IEEE Abstract SMA agorthms have recenty receved

More information

Neural network-based athletics performance prediction optimization model applied research

Neural network-based athletics performance prediction optimization model applied research Avaabe onne www.jocpr.com Journa of Chemca and Pharmaceutca Research, 04, 6(6):8-5 Research Artce ISSN : 0975-784 CODEN(USA) : JCPRC5 Neura networ-based athetcs performance predcton optmzaton mode apped

More information

Predicting Model of Traffic Volume Based on Grey-Markov

Predicting Model of Traffic Volume Based on Grey-Markov Vo. No. Modern Apped Scence Predctng Mode of Traffc Voume Based on Grey-Marov Ynpeng Zhang Zhengzhou Muncpa Engneerng Desgn & Research Insttute Zhengzhou 5005 Chna Abstract Grey-marov forecastng mode of

More information

Research on Complex Networks Control Based on Fuzzy Integral Sliding Theory

Research on Complex Networks Control Based on Fuzzy Integral Sliding Theory Advanced Scence and Technoogy Letters Vo.83 (ISA 205), pp.60-65 http://dx.do.org/0.4257/ast.205.83.2 Research on Compex etworks Contro Based on Fuzzy Integra Sdng Theory Dongsheng Yang, Bngqng L, 2, He

More information

Optimum Selection Combining for M-QAM on Fading Channels

Optimum Selection Combining for M-QAM on Fading Channels Optmum Seecton Combnng for M-QAM on Fadng Channes M. Surendra Raju, Ramesh Annavajjaa and A. Chockangam Insca Semconductors Inda Pvt. Ltd, Bangaore-56000, Inda Department of ECE, Unversty of Caforna, San

More information

Achieving Optimal Throughput Utility and Low Delay with CSMA-like Algorithms: A Virtual Multi-Channel Approach

Achieving Optimal Throughput Utility and Low Delay with CSMA-like Algorithms: A Virtual Multi-Channel Approach IEEE/AM TRANSATIONS ON NETWORKING, VOL. X, NO. XX, XXXXXXX 20X Achevng Optma Throughput Utty and Low Deay wth SMA-ke Agorthms: A Vrtua Mut-hanne Approach Po-Ka Huang, Student Member, IEEE, and Xaojun Ln,

More information

A parametric Linear Programming Model Describing Bandwidth Sharing Policies for ABR Traffic

A parametric Linear Programming Model Describing Bandwidth Sharing Policies for ABR Traffic parametrc Lnear Programmng Mode Descrbng Bandwdth Sharng Poces for BR Traffc I. Moschoos, M. Logothets and G. Kokknaks Wre ommuncatons Laboratory, Dept. of Eectrca & omputer Engneerng, Unversty of Patras,

More information

Resource Allocation with a Budget Constraint for Computing Independent Tasks in the Cloud

Resource Allocation with a Budget Constraint for Computing Independent Tasks in the Cloud Resource Allocaton wth a Budget Constrant for Computng Independent Tasks n the Cloud Wemng Sh and Bo Hong School of Electrcal and Computer Engneerng Georga Insttute of Technology, USA 2nd IEEE Internatonal

More information

Lower Bounding Procedures for the Single Allocation Hub Location Problem

Lower Bounding Procedures for the Single Allocation Hub Location Problem Lower Boundng Procedures for the Snge Aocaton Hub Locaton Probem Borzou Rostam 1,2 Chrstoph Buchhem 1,4 Fautät für Mathemat, TU Dortmund, Germany J. Faban Meer 1,3 Uwe Causen 1 Insttute of Transport Logstcs,

More information

Image Classification Using EM And JE algorithms

Image Classification Using EM And JE algorithms Machne earnng project report Fa, 2 Xaojn Sh, jennfer@soe Image Cassfcaton Usng EM And JE agorthms Xaojn Sh Department of Computer Engneerng, Unversty of Caforna, Santa Cruz, CA, 9564 jennfer@soe.ucsc.edu

More information

Application of Particle Swarm Optimization to Economic Dispatch Problem: Advantages and Disadvantages

Application of Particle Swarm Optimization to Economic Dispatch Problem: Advantages and Disadvantages Appcaton of Partce Swarm Optmzaton to Economc Dspatch Probem: Advantages and Dsadvantages Kwang Y. Lee, Feow, IEEE, and Jong-Bae Par, Member, IEEE Abstract--Ths paper summarzes the state-of-art partce

More information

A General Column Generation Algorithm Applied to System Reliability Optimization Problems

A General Column Generation Algorithm Applied to System Reliability Optimization Problems A Genera Coumn Generaton Agorthm Apped to System Reabty Optmzaton Probems Lea Za, Davd W. Cot, Department of Industra and Systems Engneerng, Rutgers Unversty, Pscataway, J 08854, USA Abstract A genera

More information

[WAVES] 1. Waves and wave forces. Definition of waves

[WAVES] 1. Waves and wave forces. Definition of waves 1. Waves and forces Defnton of s In the smuatons on ong-crested s are consdered. The drecton of these s (μ) s defned as sketched beow n the goba co-ordnate sstem: North West East South The eevaton can

More information

Numerical Investigation of Power Tunability in Two-Section QD Superluminescent Diodes

Numerical Investigation of Power Tunability in Two-Section QD Superluminescent Diodes Numerca Investgaton of Power Tunabty n Two-Secton QD Superumnescent Dodes Matta Rossett Paoo Bardea Ivo Montrosset POLITECNICO DI TORINO DELEN Summary 1. A smpfed mode for QD Super Lumnescent Dodes (SLD)

More information

Interference Alignment and Degrees of Freedom Region of Cellular Sigma Channel

Interference Alignment and Degrees of Freedom Region of Cellular Sigma Channel 2011 IEEE Internatona Symposum on Informaton Theory Proceedngs Interference Agnment and Degrees of Freedom Regon of Ceuar Sgma Channe Huaru Yn 1 Le Ke 2 Zhengdao Wang 2 1 WINLAB Dept of EEIS Unv. of Sc.

More information

Downlink Power Allocation for CoMP-NOMA in Multi-Cell Networks

Downlink Power Allocation for CoMP-NOMA in Multi-Cell Networks Downn Power Aocaton for CoMP-NOMA n Mut-Ce Networs Md Shpon A, Eram Hossan, Arafat A-Dwe, and Dong In Km arxv:80.0498v [eess.sp] 6 Dec 207 Abstract Ths wor consders the probem of dynamc power aocaton n

More information

Lecture Notes on Linear Regression

Lecture Notes on Linear Regression Lecture Notes on Lnear Regresson Feng L fl@sdueducn Shandong Unversty, Chna Lnear Regresson Problem In regresson problem, we am at predct a contnuous target value gven an nput feature vector We assume

More information

Distributed Moving Horizon State Estimation of Nonlinear Systems. Jing Zhang

Distributed Moving Horizon State Estimation of Nonlinear Systems. Jing Zhang Dstrbuted Movng Horzon State Estmaton of Nonnear Systems by Jng Zhang A thess submtted n parta fufment of the requrements for the degree of Master of Scence n Chemca Engneerng Department of Chemca and

More information

Cyclic Codes BCH Codes

Cyclic Codes BCH Codes Cycc Codes BCH Codes Gaos Feds GF m A Gaos fed of m eements can be obtaned usng the symbos 0,, á, and the eements beng 0,, á, á, á 3 m,... so that fed F* s cosed under mutpcaton wth m eements. The operator

More information

3.1 Expectation of Functions of Several Random Variables. )' be a k-dimensional discrete or continuous random vector, with joint PMF p (, E X E X1 E X

3.1 Expectation of Functions of Several Random Variables. )' be a k-dimensional discrete or continuous random vector, with joint PMF p (, E X E X1 E X Statstcs 1: Probablty Theory II 37 3 EPECTATION OF SEVERAL RANDOM VARIABLES As n Probablty Theory I, the nterest n most stuatons les not on the actual dstrbuton of a random vector, but rather on a number

More information

Module 3 LOSSY IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur

Module 3 LOSSY IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur Module 3 LOSSY IMAGE COMPRESSION SYSTEMS Verson ECE IIT, Kharagpur Lesson 6 Theory of Quantzaton Verson ECE IIT, Kharagpur Instructonal Objectves At the end of ths lesson, the students should be able to:

More information

Polite Water-filling for Weighted Sum-rate Maximization in MIMO B-MAC Networks under. Multiple Linear Constraints

Polite Water-filling for Weighted Sum-rate Maximization in MIMO B-MAC Networks under. Multiple Linear Constraints 2011 IEEE Internatona Symposum on Informaton Theory Proceedngs Pote Water-fng for Weghted Sum-rate Maxmzaton n MIMO B-MAC Networks under Mutpe near Constrants An u 1, Youjan u 2, Vncent K. N. au 3, Hage

More information

ON AUTOMATIC CONTINUITY OF DERIVATIONS FOR BANACH ALGEBRAS WITH INVOLUTION

ON AUTOMATIC CONTINUITY OF DERIVATIONS FOR BANACH ALGEBRAS WITH INVOLUTION European Journa of Mathematcs and Computer Scence Vo. No. 1, 2017 ON AUTOMATC CONTNUTY OF DERVATONS FOR BANACH ALGEBRAS WTH NVOLUTON Mohamed BELAM & Youssef T DL MATC Laboratory Hassan Unversty MORO CCO

More information

Structure and Drive Paul A. Jensen Copyright July 20, 2003

Structure and Drive Paul A. Jensen Copyright July 20, 2003 Structure and Drve Paul A. Jensen Copyrght July 20, 2003 A system s made up of several operatons wth flow passng between them. The structure of the system descrbes the flow paths from nputs to outputs.

More information

LECTURE 21 Mohr s Method for Calculation of General Displacements. 1 The Reciprocal Theorem

LECTURE 21 Mohr s Method for Calculation of General Displacements. 1 The Reciprocal Theorem V. DEMENKO MECHANICS OF MATERIALS 05 LECTURE Mohr s Method for Cacuaton of Genera Dspacements The Recproca Theorem The recproca theorem s one of the genera theorems of strength of materas. It foows drect

More information

On Uplink-Downlink Sum-MSE Duality of Multi-hop MIMO Relay Channel

On Uplink-Downlink Sum-MSE Duality of Multi-hop MIMO Relay Channel On Upn-Downn Sum-MSE Duat of Mut-hop MIMO Rea Channe A Cagata Cr, Muhammad R. A. handaer, Yue Rong and Yngbo ua Department of Eectrca Engneerng, Unverst of Caforna Rversde, Rversde, CA, 95 Centre for Wreess

More information

Inthem-machine flow shop problem, a set of jobs, each

Inthem-machine flow shop problem, a set of jobs, each THE ASYMPTOTIC OPTIMALITY OF THE SPT RULE FOR THE FLOW SHOP MEAN COMPLETION TIME PROBLEM PHILIP KAMINSKY Industra Engneerng and Operatons Research, Unversty of Caforna, Bereey, Caforna 9470, amnsy@eor.bereey.edu

More information

Boundary Value Problems. Lecture Objectives. Ch. 27

Boundary Value Problems. Lecture Objectives. Ch. 27 Boundar Vaue Probes Ch. 7 Lecture Obectves o understand the dfference between an nta vaue and boundar vaue ODE o be abe to understand when and how to app the shootng ethod and FD ethod. o understand what

More information

Quantum Runge-Lenz Vector and the Hydrogen Atom, the hidden SO(4) symmetry

Quantum Runge-Lenz Vector and the Hydrogen Atom, the hidden SO(4) symmetry Quantum Runge-Lenz ector and the Hydrogen Atom, the hdden SO(4) symmetry Pasca Szrftgser and Edgardo S. Cheb-Terrab () Laboratore PhLAM, UMR CNRS 85, Unversté Le, F-59655, France () Mapesoft Let's consder

More information

3. Stress-strain relationships of a composite layer

3. Stress-strain relationships of a composite layer OM PO I O U P U N I V I Y O F W N ompostes ourse 8-9 Unversty of wente ng. &ech... tress-stran reatonshps of a composte ayer - Laurent Warnet & emo Aerman.. tress-stran reatonshps of a composte ayer Introducton

More information

Assortment Optimization under MNL

Assortment Optimization under MNL Assortment Optmzaton under MNL Haotan Song Aprl 30, 2017 1 Introducton The assortment optmzaton problem ams to fnd the revenue-maxmzng assortment of products to offer when the prces of products are fxed.

More information

L-Edge Chromatic Number Of A Graph

L-Edge Chromatic Number Of A Graph IJISET - Internatona Journa of Innovatve Scence Engneerng & Technoogy Vo. 3 Issue 3 March 06. ISSN 348 7968 L-Edge Chromatc Number Of A Graph Dr.R.B.Gnana Joth Assocate Professor of Mathematcs V.V.Vannaperuma

More information

2E Pattern Recognition Solutions to Introduction to Pattern Recognition, Chapter 2: Bayesian pattern classification

2E Pattern Recognition Solutions to Introduction to Pattern Recognition, Chapter 2: Bayesian pattern classification E395 - Pattern Recognton Solutons to Introducton to Pattern Recognton, Chapter : Bayesan pattern classfcaton Preface Ths document s a soluton manual for selected exercses from Introducton to Pattern Recognton

More information

Associative Memories

Associative Memories Assocatve Memores We consder now modes for unsupervsed earnng probems, caed auto-assocaton probems. Assocaton s the task of mappng patterns to patterns. In an assocatve memory the stmuus of an ncompete

More information

Problem Set 9 Solutions

Problem Set 9 Solutions Desgn and Analyss of Algorthms May 4, 2015 Massachusetts Insttute of Technology 6.046J/18.410J Profs. Erk Demane, Srn Devadas, and Nancy Lynch Problem Set 9 Solutons Problem Set 9 Solutons Ths problem

More information

Chapter 6. Rotations and Tensors

Chapter 6. Rotations and Tensors Vector Spaces n Physcs 8/6/5 Chapter 6. Rotatons and ensors here s a speca knd of near transformaton whch s used to transforms coordnates from one set of axes to another set of axes (wth the same orgn).

More information

Lower bounds for the Crossing Number of the Cartesian Product of a Vertex-transitive Graph with a Cycle

Lower bounds for the Crossing Number of the Cartesian Product of a Vertex-transitive Graph with a Cycle Lower bounds for the Crossng Number of the Cartesan Product of a Vertex-transtve Graph wth a Cyce Junho Won MIT-PRIMES December 4, 013 Abstract. The mnmum number of crossngs for a drawngs of a gven graph

More information

DISTRIBUTED PROCESSING OVER ADAPTIVE NETWORKS. Cassio G. Lopes and Ali H. Sayed

DISTRIBUTED PROCESSING OVER ADAPTIVE NETWORKS. Cassio G. Lopes and Ali H. Sayed DISTRIBUTED PROCESSIG OVER ADAPTIVE ETWORKS Casso G Lopes and A H Sayed Department of Eectrca Engneerng Unversty of Caforna Los Angees, CA, 995 Ema: {casso, sayed@eeucaedu ABSTRACT Dstrbuted adaptve agorthms

More information

I. INTRODUCTION WIRELESS sensing and control systems have received

I. INTRODUCTION WIRELESS sensing and control systems have received IEEE TRANSACTIONS ON CONTROL SYSTEMS TECHNOLOGY, VOL. 23, NO. 6, NOVEMBER 2015 2167 Mutpe-Loop Sef-Trggered Mode Predctve Contro for Network Schedung and Contro Erk Henrksson, Dane E. Quevedo, Senor Member,

More information

A MIN-MAX REGRET ROBUST OPTIMIZATION APPROACH FOR LARGE SCALE FULL FACTORIAL SCENARIO DESIGN OF DATA UNCERTAINTY

A MIN-MAX REGRET ROBUST OPTIMIZATION APPROACH FOR LARGE SCALE FULL FACTORIAL SCENARIO DESIGN OF DATA UNCERTAINTY A MIN-MAX REGRET ROBST OPTIMIZATION APPROACH FOR ARGE SCAE F FACTORIA SCENARIO DESIGN OF DATA NCERTAINTY Travat Assavapokee Department of Industra Engneerng, nversty of Houston, Houston, Texas 7704-4008,

More information

The Application of BP Neural Network principal component analysis in the Forecasting the Road Traffic Accident

The Application of BP Neural Network principal component analysis in the Forecasting the Road Traffic Accident ICTCT Extra Workshop, Bejng Proceedngs The Appcaton of BP Neura Network prncpa component anayss n Forecastng Road Traffc Accdent He Mng, GuoXucheng &LuGuangmng Transportaton Coege of Souast Unversty 07

More information

Outline. Communication. Bellman Ford Algorithm. Bellman Ford Example. Bellman Ford Shortest Path [1]

Outline. Communication. Bellman Ford Algorithm. Bellman Ford Example. Bellman Ford Shortest Path [1] DYNAMIC SHORTEST PATH SEARCH AND SYNCHRONIZED TASK SWITCHING Jay Wagenpfel, Adran Trachte 2 Outlne Shortest Communcaton Path Searchng Bellmann Ford algorthm Algorthm for dynamc case Modfcatons to our algorthm

More information

Chapter - 2. Distribution System Power Flow Analysis

Chapter - 2. Distribution System Power Flow Analysis Chapter - 2 Dstrbuton System Power Flow Analyss CHAPTER - 2 Radal Dstrbuton System Load Flow 2.1 Introducton Load flow s an mportant tool [66] for analyzng electrcal power system network performance. Load

More information

Quantitative Evaluation Method of Each Generation Margin for Power System Planning

Quantitative Evaluation Method of Each Generation Margin for Power System Planning 1 Quanttatve Evauaton Method of Each Generaton Margn for Poer System Pannng SU Su Department of Eectrca Engneerng, Tohoku Unversty, Japan moden25@ececetohokuacjp Abstract Increasng effcency of poer pants

More information

A Derivative-Free Algorithm for Bound Constrained Optimization

A Derivative-Free Algorithm for Bound Constrained Optimization Computatona Optmzaton and Appcatons, 21, 119 142, 2002 c 2002 Kuwer Academc Pubshers. Manufactured n The Netherands. A Dervatve-Free Agorthm for Bound Constraned Optmzaton STEFANO LUCIDI ucd@ds.unroma.t

More information

Delay tomography for large scale networks

Delay tomography for large scale networks Deay tomography for arge scae networks MENG-FU SHIH ALFRED O. HERO III Communcatons and Sgna Processng Laboratory Eectrca Engneerng and Computer Scence Department Unversty of Mchgan, 30 Bea. Ave., Ann

More information

LOW BIAS INTEGRATED PATH ESTIMATORS. James M. Calvin

LOW BIAS INTEGRATED PATH ESTIMATORS. James M. Calvin Proceedngs of the 007 Wnter Smulaton Conference S G Henderson, B Bller, M-H Hseh, J Shortle, J D Tew, and R R Barton, eds LOW BIAS INTEGRATED PATH ESTIMATORS James M Calvn Department of Computer Scence

More information

Real-Time Systems. Multiprocessor scheduling. Multiprocessor scheduling. Multiprocessor scheduling

Real-Time Systems. Multiprocessor scheduling. Multiprocessor scheduling. Multiprocessor scheduling Real-Tme Systems Multprocessor schedulng Specfcaton Implementaton Verfcaton Multprocessor schedulng -- -- Global schedulng How are tasks assgned to processors? Statc assgnment The processor(s) used for

More information

Example: Suppose we want to build a classifier that recognizes WebPages of graduate students.

Example: Suppose we want to build a classifier that recognizes WebPages of graduate students. Exampe: Suppose we want to bud a cassfer that recognzes WebPages of graduate students. How can we fnd tranng data? We can browse the web and coect a sampe of WebPages of graduate students of varous unverstes.

More information

QUARTERLY OF APPLIED MATHEMATICS

QUARTERLY OF APPLIED MATHEMATICS QUARTERLY OF APPLIED MATHEMATICS Voume XLI October 983 Number 3 DIAKOPTICS OR TEARING-A MATHEMATICAL APPROACH* By P. W. AITCHISON Unversty of Mantoba Abstract. The method of dakoptcs or tearng was ntroduced

More information

Uncertainty Specification and Propagation for Loss Estimation Using FOSM Methods

Uncertainty Specification and Propagation for Loss Estimation Using FOSM Methods Uncertanty Specfcaton and Propagaton for Loss Estmaton Usng FOSM Methods J.W. Baer and C.A. Corne Dept. of Cv and Envronmenta Engneerng, Stanford Unversty, Stanford, CA 94305-400 Keywords: Sesmc, oss estmaton,

More information

NP-Completeness : Proofs

NP-Completeness : Proofs NP-Completeness : Proofs Proof Methods A method to show a decson problem Π NP-complete s as follows. (1) Show Π NP. (2) Choose an NP-complete problem Π. (3) Show Π Π. A method to show an optmzaton problem

More information

REDUCTION OF CORRELATION COMPUTATION IN THE PERMUTATION OF THE FREQUENCY DOMAIN ICA BY SELECTING DOAS ESTIMATED IN SUBARRAYS

REDUCTION OF CORRELATION COMPUTATION IN THE PERMUTATION OF THE FREQUENCY DOMAIN ICA BY SELECTING DOAS ESTIMATED IN SUBARRAYS 15th European Sgna Processng Conference (EUSIPCO 27), Poznan, Poand, September 3-7, 27, copyrght by EURASIP REDUCTION OF CORRELATION COMPUTATION IN THE PERMUTATION OF THE FREQUENCY DOMAIN ICA BY SELECTING

More information

Minimisation of the Average Response Time in a Cluster of Servers

Minimisation of the Average Response Time in a Cluster of Servers Mnmsaton of the Average Response Tme n a Cluster of Servers Valery Naumov Abstract: In ths paper, we consder task assgnment problem n a cluster of servers. We show that optmal statc task assgnment s tantamount

More information

ECE559VV Project Report

ECE559VV Project Report ECE559VV Project Report (Supplementary Notes Loc Xuan Bu I. MAX SUM-RATE SCHEDULING: THE UPLINK CASE We have seen (n the presentaton that, for downlnk (broadcast channels, the strategy maxmzng the sum-rate

More information

A FAST HEURISTIC FOR TASKS ASSIGNMENT IN MANYCORE SYSTEMS WITH VOLTAGE-FREQUENCY ISLANDS

A FAST HEURISTIC FOR TASKS ASSIGNMENT IN MANYCORE SYSTEMS WITH VOLTAGE-FREQUENCY ISLANDS Shervn Haamn A FAST HEURISTIC FOR TASKS ASSIGNMENT IN MANYCORE SYSTEMS WITH VOLTAGE-FREQUENCY ISLANDS INTRODUCTION Increasng computatons n applcatons has led to faster processng. o Use more cores n a chp

More information

On the Power Function of the Likelihood Ratio Test for MANOVA

On the Power Function of the Likelihood Ratio Test for MANOVA Journa of Mutvarate Anayss 8, 416 41 (00) do:10.1006/jmva.001.036 On the Power Functon of the Lkehood Rato Test for MANOVA Dua Kumar Bhaumk Unversty of South Aabama and Unversty of Inos at Chcago and Sanat

More information

The line method combined with spectral chebyshev for space-time fractional diffusion equation

The line method combined with spectral chebyshev for space-time fractional diffusion equation Apped and Computatona Mathematcs 014; 3(6): 330-336 Pubshed onne December 31, 014 (http://www.scencepubshnggroup.com/j/acm) do: 10.1164/j.acm.0140306.17 ISS: 3-5605 (Prnt); ISS: 3-5613 (Onne) The ne method

More information

ENERGY EFFICIENT WIRELESS TRANSMISSION OF MPEG-4 FINE GRANULAR SCALABLE VIDEO

ENERGY EFFICIENT WIRELESS TRANSMISSION OF MPEG-4 FINE GRANULAR SCALABLE VIDEO NGY FFICINT WISS TANSMISSION OF MG-4 FIN GANUA SCAA VIDO Crstna. Costa, Yftach senberg 2, Fan Zha 2, Aggeos K. Katsaggeos 2 DIT, Unversty of Trento -Va Sommarve 4, I-38 Trento (ITAY) - crstna.costa@ng.untn.t

More information

Research Article H Estimates for Discrete-Time Markovian Jump Linear Systems

Research Article H Estimates for Discrete-Time Markovian Jump Linear Systems Mathematca Probems n Engneerng Voume 213 Artce ID 945342 7 pages http://dxdoorg/11155/213/945342 Research Artce H Estmates for Dscrete-Tme Markovan Jump Lnear Systems Marco H Terra 1 Gdson Jesus 2 and

More information

n-step cycle inequalities: facets for continuous n-mixing set and strong cuts for multi-module capacitated lot-sizing problem

n-step cycle inequalities: facets for continuous n-mixing set and strong cuts for multi-module capacitated lot-sizing problem n-step cyce nequates: facets for contnuous n-mxng set and strong cuts for mut-modue capactated ot-szng probem Mansh Bansa and Kavash Kanfar Department of Industra and Systems Engneerng, Texas A&M Unversty,

More information

Simultaneous Optimization of Berth Allocation, Quay Crane Assignment and Quay Crane Scheduling Problems in Container Terminals

Simultaneous Optimization of Berth Allocation, Quay Crane Assignment and Quay Crane Scheduling Problems in Container Terminals Smultaneous Optmzaton of Berth Allocaton, Quay Crane Assgnment and Quay Crane Schedulng Problems n Contaner Termnals Necat Aras, Yavuz Türkoğulları, Z. Caner Taşkın, Kuban Altınel Abstract In ths work,

More information

ANSWERS. Problem 1. and the moment generating function (mgf) by. defined for any real t. Use this to show that E( U) var( U)

ANSWERS. Problem 1. and the moment generating function (mgf) by. defined for any real t. Use this to show that E( U) var( U) Econ 413 Exam 13 H ANSWERS Settet er nndelt 9 deloppgaver, A,B,C, som alle anbefales å telle lkt for å gøre det ltt lettere å stå. Svar er gtt . Unfortunately, there s a prntng error n the hnt of

More information

Adaptive Divisible Load Scheduling in Computational Grids Luis de la Torre 1 Héctor de la Torre 2 1

Adaptive Divisible Load Scheduling in Computational Grids Luis de la Torre 1 Héctor de la Torre 2 1 Adaptve Dvsbe Load Schedung n Computatona Grds Lus de a Torre Héctor de a Torre 2 Scence and Technoogy Schoo, Unversdad Metropotana, San Juan, PR 2 Schoo of Engneerng, Turabo Unversty, Gurabo, PR Ana G.

More information

Single-Source/Sink Network Error Correction Is as Hard as Multiple-Unicast

Single-Source/Sink Network Error Correction Is as Hard as Multiple-Unicast Snge-Source/Snk Network Error Correcton Is as Hard as Mutpe-Uncast Wentao Huang and Tracey Ho Department of Eectrca Engneerng Caforna Insttute of Technoogy Pasadena, CA {whuang,tho}@catech.edu Mchae Langberg

More information

CSci 6974 and ECSE 6966 Math. Tech. for Vision, Graphics and Robotics Lecture 21, April 17, 2006 Estimating A Plane Homography

CSci 6974 and ECSE 6966 Math. Tech. for Vision, Graphics and Robotics Lecture 21, April 17, 2006 Estimating A Plane Homography CSc 6974 and ECSE 6966 Math. Tech. for Vson, Graphcs and Robotcs Lecture 21, Aprl 17, 2006 Estmatng A Plane Homography Overvew We contnue wth a dscusson of the major ssues, usng estmaton of plane projectve

More information

COXREG. Estimation (1)

COXREG. Estimation (1) COXREG Cox (972) frst suggested the modes n whch factors reated to fetme have a mutpcatve effect on the hazard functon. These modes are caed proportona hazards (PH) modes. Under the proportona hazards

More information

Week3, Chapter 4. Position and Displacement. Motion in Two Dimensions. Instantaneous Velocity. Average Velocity

Week3, Chapter 4. Position and Displacement. Motion in Two Dimensions. Instantaneous Velocity. Average Velocity Week3, Chapter 4 Moton n Two Dmensons Lecture Quz A partcle confned to moton along the x axs moves wth constant acceleraton from x =.0 m to x = 8.0 m durng a 1-s tme nterval. The velocty of the partcle

More information

Multispectral Remote Sensing Image Classification Algorithm Based on Rough Set Theory

Multispectral Remote Sensing Image Classification Algorithm Based on Rough Set Theory Proceedngs of the 2009 IEEE Internatona Conference on Systems Man and Cybernetcs San Antono TX USA - October 2009 Mutspectra Remote Sensng Image Cassfcaton Agorthm Based on Rough Set Theory Yng Wang Xaoyun

More information

CABLE STRUCTURE WITH LOAD-ADAPTING GEOMETRY

CABLE STRUCTURE WITH LOAD-ADAPTING GEOMETRY Compostes n Constructon Thrd Internatona Conerence Lon, France, Ju 3, CABLE STRUCTURE WITH LOAD-ADAPTING GEOMETRY A.S. Jüch, J.F. Caron and O. Bavere Insttut Naver - Lam, ENPC-LCPC 6 et, avenue Base Pasca

More information

Subgradient Methods and Consensus Algorithms for Solving Convex Optimization Problems

Subgradient Methods and Consensus Algorithms for Solving Convex Optimization Problems Proceedngs of the 47th IEEE Conference on Decson and Contro Cancun, Mexco, Dec. 9-11, 2008 Subgradent Methods and Consensus Agorthms for Sovng Convex Optmzaton Probems Björn Johansson, Tamás Kevczy, Mae

More information

Greyworld White Balancing with Low Computation Cost for On- Board Video Capturing

Greyworld White Balancing with Low Computation Cost for On- Board Video Capturing reyword Whte aancng wth Low Computaton Cost for On- oard Vdeo Capturng Peng Wu Yuxn Zoe) Lu Hewett-Packard Laboratores Hewett-Packard Co. Pao Ato CA 94304 USA Abstract Whte baancng s a process commony

More information

Reservation Policies for Revenue Maximization from Secondary Spectrum Access in Cellular Networks

Reservation Policies for Revenue Maximization from Secondary Spectrum Access in Cellular Networks Reservaton Poces for Revenue Maxmzaton from Secondary Spectrum Access n Ceuar Networks Ashraf A Daoud, Murat Aanya, and Davd Starobnsk Department of Eectrca and Computer Engneerng Boston Unversty, Boston,

More information

Winter 2008 CS567 Stochastic Linear/Integer Programming Guest Lecturer: Xu, Huan

Winter 2008 CS567 Stochastic Linear/Integer Programming Guest Lecturer: Xu, Huan Wnter 2008 CS567 Stochastc Lnear/Integer Programmng Guest Lecturer: Xu, Huan Class 2: More Modelng Examples 1 Capacty Expanson Capacty expanson models optmal choces of the tmng and levels of nvestments

More information

Dynamic Analysis Of An Off-Road Vehicle Frame

Dynamic Analysis Of An Off-Road Vehicle Frame Proceedngs of the 8th WSEAS Int. Conf. on NON-LINEAR ANALYSIS, NON-LINEAR SYSTEMS AND CHAOS Dnamc Anass Of An Off-Road Vehce Frame ŞTEFAN TABACU, NICOLAE DORU STĂNESCU, ION TABACU Automotve Department,

More information

Correspondence. Performance Evaluation for MAP State Estimate Fusion I. INTRODUCTION

Correspondence. Performance Evaluation for MAP State Estimate Fusion I. INTRODUCTION Correspondence Performance Evauaton for MAP State Estmate Fuson Ths paper presents a quanttatve performance evauaton method for the maxmum a posteror (MAP) state estmate fuson agorthm. Under dea condtons

More information

Reliability Sensitivity Algorithm Based on Stratified Importance Sampling Method for Multiple Failure Modes Systems

Reliability Sensitivity Algorithm Based on Stratified Importance Sampling Method for Multiple Failure Modes Systems Chnese Journa o Aeronautcs 3(010) 660-669 Chnese Journa o Aeronautcs www.esever.com/ocate/ca Reabty Senstvty Agorthm Based on Strated Importance Sampng Method or Mutpe aure Modes Systems Zhang eng a, u

More information

MULTIVARIABLE FUZZY CONTROL WITH ITS APPLICATIONS IN MULTI EVAPORATOR REFRIGERATION SYSTEMS

MULTIVARIABLE FUZZY CONTROL WITH ITS APPLICATIONS IN MULTI EVAPORATOR REFRIGERATION SYSTEMS MULTIVARIABLE FUZZY CONTROL WITH I APPLICATIONS IN MULTI EVAPORATOR REFRIGERATION SYSTEMS LIAO QIANFANG Schoo of Eectrca and Eectronc Engneerng A thess submtted to the Nanyang Technoogca Unversty n parta

More information

REAL-TIME IMPACT FORCE IDENTIFICATION OF CFRP LAMINATED PLATES USING SOUND WAVES

REAL-TIME IMPACT FORCE IDENTIFICATION OF CFRP LAMINATED PLATES USING SOUND WAVES 8 TH INTERNATIONAL CONERENCE ON COMPOSITE MATERIALS REAL-TIME IMPACT ORCE IDENTIICATION O CRP LAMINATED PLATES USING SOUND WAVES S. Atobe *, H. Kobayash, N. Hu 3 and H. ukunaga Department of Aerospace

More information

Composite Hypotheses testing

Composite Hypotheses testing Composte ypotheses testng In many hypothess testng problems there are many possble dstrbutons that can occur under each of the hypotheses. The output of the source s a set of parameters (ponts n a parameter

More information

A Simple Congestion-Aware Algorithm for Load Balancing in Datacenter Networks

A Simple Congestion-Aware Algorithm for Load Balancing in Datacenter Networks IEEE/ACM TRANSACTIONS ON NETWORKING A Smpe Congeston-Aware Agorthm for Load Baancng n Datacenter Networs Mehrnoosh Shafee, Student Member, IEEE, and Javad Ghader, Member, IEEE Abstract We study the probem

More information

Xin Li Department of Information Systems, College of Business, City University of Hong Kong, Hong Kong, CHINA

Xin Li Department of Information Systems, College of Business, City University of Hong Kong, Hong Kong, CHINA RESEARCH ARTICLE MOELING FIXE OS BETTING FOR FUTURE EVENT PREICTION Weyun Chen eartment of Educatona Informaton Technoogy, Facuty of Educaton, East Chna Norma Unversty, Shangha, CHINA {weyun.chen@qq.com}

More information

A CONSTRAINT GENERATION INTEGER PROGRAMMING APPROACH TO INFORMATION THEORETIC SENSOR RESOURCE MANAGEMENT

A CONSTRAINT GENERATION INTEGER PROGRAMMING APPROACH TO INFORMATION THEORETIC SENSOR RESOURCE MANAGEMENT A CONSTRAINT GENERATION INTEGER PROGRAMMING APPROACH TO INFORMATION THEORETIC SENSOR RESOURCE MANAGEMENT Jason L. Wams, 1, John W. Fsher III, Aan S. Wsy 1, 1 Laboratory for Informaton and Decson Systems

More information

Development of whole CORe Thermal Hydraulic analysis code CORTH Pan JunJie, Tang QiFen, Chai XiaoMing, Lu Wei, Liu Dong

Development of whole CORe Thermal Hydraulic analysis code CORTH Pan JunJie, Tang QiFen, Chai XiaoMing, Lu Wei, Liu Dong Deveopment of whoe CORe Therma Hydrauc anayss code CORTH Pan JunJe, Tang QFen, Cha XaoMng, Lu We, Lu Dong cence and technoogy on reactor system desgn technoogy, Nucear Power Insttute of Chna, Chengdu,

More information

A DIMENSION-REDUCTION METHOD FOR STOCHASTIC ANALYSIS SECOND-MOMENT ANALYSIS

A DIMENSION-REDUCTION METHOD FOR STOCHASTIC ANALYSIS SECOND-MOMENT ANALYSIS A DIMESIO-REDUCTIO METHOD FOR STOCHASTIC AALYSIS SECOD-MOMET AALYSIS S. Rahman Department of Mechanca Engneerng and Center for Computer-Aded Desgn The Unversty of Iowa Iowa Cty, IA 52245 June 2003 OUTLIE

More information

Integrating advanced demand models within the framework of mixed integer linear problems: A Lagrangian relaxation method for the uncapacitated

Integrating advanced demand models within the framework of mixed integer linear problems: A Lagrangian relaxation method for the uncapacitated Integratng advanced demand modes wthn the framework of mxed nteger near probems: A Lagrangan reaxaton method for the uncapactated case Mertxe Pacheco Paneque Shad Sharf Azadeh Mche Berare Bernard Gendron

More information

The Second Anti-Mathima on Game Theory

The Second Anti-Mathima on Game Theory The Second Ant-Mathma on Game Theory Ath. Kehagas December 1 2006 1 Introducton In ths note we wll examne the noton of game equlbrum for three types of games 1. 2-player 2-acton zero-sum games 2. 2-player

More information

Short-Term Load Forecasting for Electric Power Systems Using the PSO-SVR and FCM Clustering Techniques

Short-Term Load Forecasting for Electric Power Systems Using the PSO-SVR and FCM Clustering Techniques Energes 20, 4, 73-84; do:0.3390/en40073 Artce OPEN ACCESS energes ISSN 996-073 www.mdp.com/journa/energes Short-Term Load Forecastng for Eectrc Power Systems Usng the PSO-SVR and FCM Custerng Technques

More information

Appendix for An Efficient Ascending-Bid Auction for Multiple Objects: Comment For Online Publication

Appendix for An Efficient Ascending-Bid Auction for Multiple Objects: Comment For Online Publication Appendx for An Effcent Ascendng-Bd Aucton for Mutpe Objects: Comment For Onne Pubcaton Norak Okamoto The foowng counterexampe shows that sncere bddng by a bdders s not aways an ex post perfect equbrum

More information

GENERATION OF GOLD-SEQUENCES WITH APPLICATIONS TO SPREAD SPECTRUM SYSTEMS

GENERATION OF GOLD-SEQUENCES WITH APPLICATIONS TO SPREAD SPECTRUM SYSTEMS GENERATION OF GOLD-SEQUENCES WITH APPLICATIONS TO SPREAD SPECTRUM SYSTEMS F. Rodríguez Henríquez (1), Member, IEEE, N. Cruz Cortés (1), Member, IEEE, J.M. Rocha-Pérez (2) Member, IEEE. F. Amaro Sánchez

More information

Single-Facility Scheduling over Long Time Horizons by Logic-based Benders Decomposition

Single-Facility Scheduling over Long Time Horizons by Logic-based Benders Decomposition Sngle-Faclty Schedulng over Long Tme Horzons by Logc-based Benders Decomposton Elvn Coban and J. N. Hooker Tepper School of Busness, Carnege Mellon Unversty ecoban@andrew.cmu.edu, john@hooker.tepper.cmu.edu

More information

χ x B E (c) Figure 2.1.1: (a) a material particle in a body, (b) a place in space, (c) a configuration of the body

χ x B E (c) Figure 2.1.1: (a) a material particle in a body, (b) a place in space, (c) a configuration of the body Secton.. Moton.. The Materal Body and Moton hyscal materals n the real world are modeled usng an abstract mathematcal entty called a body. Ths body conssts of an nfnte number of materal partcles. Shown

More information

Optimal Guaranteed Cost Control of Linear Uncertain Systems with Input Constraints

Optimal Guaranteed Cost Control of Linear Uncertain Systems with Input Constraints Internatona Journa Optma of Contro, Guaranteed Automaton, Cost Contro and Systems, of Lnear vo Uncertan 3, no Systems 3, pp 397-4, wth Input September Constrants 5 397 Optma Guaranteed Cost Contro of Lnear

More information

The Minimum Universal Cost Flow in an Infeasible Flow Network

The Minimum Universal Cost Flow in an Infeasible Flow Network Journal of Scences, Islamc Republc of Iran 17(2): 175-180 (2006) Unversty of Tehran, ISSN 1016-1104 http://jscencesutacr The Mnmum Unversal Cost Flow n an Infeasble Flow Network H Saleh Fathabad * M Bagheran

More information

4 Analysis of Variance (ANOVA) 5 ANOVA. 5.1 Introduction. 5.2 Fixed Effects ANOVA

4 Analysis of Variance (ANOVA) 5 ANOVA. 5.1 Introduction. 5.2 Fixed Effects ANOVA 4 Analyss of Varance (ANOVA) 5 ANOVA 51 Introducton ANOVA ANOVA s a way to estmate and test the means of multple populatons We wll start wth one-way ANOVA If the populatons ncluded n the study are selected

More information

Note 2. Ling fong Li. 1 Klein Gordon Equation Probablity interpretation Solutions to Klein-Gordon Equation... 2

Note 2. Ling fong Li. 1 Klein Gordon Equation Probablity interpretation Solutions to Klein-Gordon Equation... 2 Note 2 Lng fong L Contents Ken Gordon Equaton. Probabty nterpretaton......................................2 Soutons to Ken-Gordon Equaton............................... 2 2 Drac Equaton 3 2. Probabty nterpretaton.....................................

More information

SIMULTANEOUS wireless information and power transfer. Joint Optimization of Power and Data Transfer in Multiuser MIMO Systems

SIMULTANEOUS wireless information and power transfer. Joint Optimization of Power and Data Transfer in Multiuser MIMO Systems Jont Optmzaton of Power and Data ransfer n Mutuser MIMO Systems Javer Rubo, Antono Pascua-Iserte, Dane P. Paomar, and Andrea Godsmth Unverstat Potècnca de Cataunya UPC, Barceona, Span ong Kong Unversty

More information

Deriving the Dual. Prof. Bennett Math of Data Science 1/13/06

Deriving the Dual. Prof. Bennett Math of Data Science 1/13/06 Dervng the Dua Prof. Bennett Math of Data Scence /3/06 Outne Ntty Grtty for SVM Revew Rdge Regresson LS-SVM=KRR Dua Dervaton Bas Issue Summary Ntty Grtty Need Dua of w, b, z w 2 2 mn st. ( x w ) = C z

More information

Nested case-control and case-cohort studies

Nested case-control and case-cohort studies Outne: Nested case-contro and case-cohort studes Ørnuf Borgan Department of Mathematcs Unversty of Oso NORBIS course Unversty of Oso 4-8 December 217 1 Radaton and breast cancer data Nested case contro

More information