Resource Allocation for Real-Time Tasks using Cloud Computing

Size: px
Start display at page:

Download "Resource Allocation for Real-Time Tasks using Cloud Computing"

Transcription

1 Reource Allocation for Real-Time Tak uing Cloud Computing Karthik Kumar, Jing Feng, Yamini Nimmagadda, and Yung-Hiang Lu School of Electrical and Computer Engineering, Purdue Univerity, Wet Lafayette, IN, Abtract Thi paper preent a method to allocate reource for real-time tak uing the Infratructure a a Service model offered by cloud computing. Real-time tak have to be completed before deadline, and cloud computing offer election of reource with different peed and cot. In cloud computing, reource allocation can be caled up baed on the requirement; thi i called elaticity and i the key difference from exiting multiproceor tak allocation. Scalable reource make economical allocation of reource an important problem. We analyze the problem of allocating reource for a et of realtime tak uch that the economic cot i minimized and all the deadline are met. We formulate the problem a a contrained optimization problem and propoe a polynomial-time olution to allocate reource efficiently. We compare the economic cot and performance provided by our olution with the optimal olution and an EDF (earliet deadline firt) method. We how how the cot varie baed on the ditribution of the tak. Index Term reource allocation; cloud computing; cheduling; I. INTRODUCTION Cloud computing offer a uer the ervice (called Infratructure a a Service - IaaS) of renting computing reource over the Internet. The uer can elect from different type of computing reource baed on the requirement. For example, Amazon EC2 Cloud ervice provide different option for electing reource a hown in Table I. The uer can rent an arbitrarily large number of reource: thi i called calable computing or elaticity, ince the number can be caled dynamically to meet the requirement. One et of application that can benefit from calable computing are mixed-parallel application [5], [12], [16], [18], [20]. Thee application exhibit high tak and data parallelim. Many of thee application are real-time [5], [12], [20] and require their workload to be completed before deadline; example include voice and object recognition, image and video retrieval, navigation ytem, financial ytem, and miion-critical ytem. For example, an object recognition engine [20] may be hoted on the cloud. Each tak i an object query; the object mut be recognized within a pecified time duration to be of value to the uer. Since the cloud offer calable computing, reource allocation can be caled baed on the arrival time, workload, and deadline of the tak. In cloud computing, a reource i a virtual machine that guarantee a certain level of performance to the uer. For example, Amazon EC2 cloud ervice define virtual machine with peed in compute unit ; thi provide a hardwareindependent definition of peed for the virtual machine by abtracting away variation in the underlying phyical hardware. The type and amount of virtual machine allocated determine the cot paid by the uer. Amazon cot for different virtual machine are hown in Table I. There are three different type tandard, high memory, and high CPU, including different amount of proceor, memory, and torage at different cot. Each virtual machine i rented multiple hour, and the uer Virtual Machine Compute Memory Storage Cot/ type Unit hour High Memory GB 420 GB $0.69 Standard GB 1690 GB $1.04 High CPU GB 1690 GB $1.24 TABLE I VIRTUAL MACHINE TYPES AVAILABLE IN AMAZON S EC2 CLOUD SERVICE (SIZE EXTRA LARGE, US-N CALIFORNIA). i charged a fixed cot irrepective of the virtual machine utilization within the hour. Thi motivate the need to find a cot-efficient allocation for a given et of tak. In the ret of thi paper, we ue the term reource, virtual machine (VM), and proceor interchangeably. Several reearcher have developed efficient allocation for real-time tak on multi-proceor ytem [7], [23], [13]. However, previou tudie chedule tak on a fixed number of proceor. For calable computing, the virtual machine (VM) are rented, and can be caled up to any number. Thi create ome fundamental change in the problem. Firt, it implie that if a tak i ufficiently parallelizable, it deadline can alway be met ince more VM can be allocated to complete the tak before it deadline. Thi i different from previou tudie that examine the chedulability on a given number of proceor. Second, ince the number of available VM i nearly infinite, at every time intant, there are option uing different number and type of VM, baed on their computing peed and cot. For example, to finih a tak, a uer can elect a larger number of lower, cheaper VM or a maller number of fater, more expenive VM, or a combination between them. Third, acquiring VM by rent implie a fixed charge for a given rental period. Suppoe a uer rent a VM for an hour and the tak complete before the end of the hour, the VM become available for running other tak that arrive within the hour. Thu the allocation of reource for one tak at the preent may affect the election of reource for future tak. Recent tudie on allocating cloud VM for real time tak [19], [12], [14], [3] focu on different apect like the infratructure to enable real-time tak on VM, election of VM for power management in the data center, etc. Unfortunately, none of thee tudie conider how the uer can make a cot-efficient election from a et of different VM for real-time tak. In thi paper, we develop an algorithm for allocating VM to application with real-time tak. The allocation i formulated a a contrained optimization problem. Since an exhautive earch for olution ha exponential complexity, we propoe a polynomial-time heuritic to olve the problem. We compare the cot obtained by our heuritic with the optimal olution, and an Earliet Deadline Firt (EDF-) trategy. We how the condition when our heuritic outperform the EDF trategy. We alo perform a enitivity analyi for the /11/$ IEEE

2 Paper Heterogeneou Workload Number of Proceor Proceor [9] Ye Fixed Fixed [22], [4] No Probabilitic Fixed [7], [23], [13] No Fixed Fixed Thi paper Ye Fixed Scalable TABLE II COMPARISON OF DIFFERENT STUDIES ON REAL-TIME SCHEDULING. parameter of the problem. The remainder of thi paper i organized a follow: Section II decribe the type of application that can be ued for our problem, and highlight our contribution by comparing it with related work. Section III decribe our problem formulation, and the propoed olution. Section IV decribe the evaluation of the propoed olution, and Section V conclude the paper. II. BACKGROUND AND RELATED WORK A. Scheduling Parallel Application Scheduling mixed-parallel application [16], [18] on multiproceor ytem i a known NP-Complete problem [16]. Many of thee parallel application are real-time [12], [5], [20]. The tak in thee application have two important characteritic: (1) highly parallelizable [11], [2], [16] and (2) real-time contraint (or deadline) [15], [17]. Each tak can be partitioned into maller and parallel unit. We ue p a the mallet unit of computation. For example, image retrieval may compare image with a query to find a match, and the mallet unit of computation i comparing a ingle image with the query. The tak find a match for a query image, called img, from a collection of one million image. Thi tak mut be completed within d econd. We may ue one million VM and each compare only one image. If all VM can execute imultaneouly and the torage ytem can provide a ufficient bandwidth, thi would be the fatet approach but it would alo ue the larget number of VM. Since we only need to find a match for img (ye/no), the reult can be merged very quickly in logarithmic time. Another option ue 1000 VM and each VM compare 1000 image. It i alo poible to ue a ingle VM for the one million image; thi will take much longer. The actual election of VM may be contrained by the deadline d, the peed and cot of different type of VM. In thi paper, we conider cot-efficient VM election for uch application. B. Scheduling for Multiproceor Previou tudie have conidered energy-efficient cheduling. Uniproceor cheduling cheme like Earliet-Deadline Firt (EDF) [8] and Rate Monotonic (RM) cheduling [1] are adapted to multiproceor ytem. Many reearcher [7], [4], [23], [13] conider real-time tak allocation on multiproceor ytem. Several paper have tudied Dynamic Voltage and Frequency Scaling (DVFS). We do not conider DVFS in thi paper a it i a well tudied problem. Table II how how our work differ from the current real-time multiproceor cheduling technique. C. Virtual Machine Allocation for Cloud Computing Thi include two different problem: (1) electing virtual reource by the uer and (2) mapping virtual reource to phyical reource by the ervice provider. Thi paper focue on the firt problem. Several tudie [10], [21], [6] conider cot-effective reource election for cloud ytem. The common focu of all thee work i cot-efficient VM allocation; the key difference from the above work i that we conider real-time tak. Recent tudie have been performed on allocation of VM for real time tak. Aymerich et al. [3] develop an infratructure for deploying real-time financial tak on cloud ytem. Tai et al. [19] dicu real-time partitioning of databae tak on cloud infratructure. Liu et al. [14] how how to chedule real-time tak with different utility function; however they do not conider different type of VM. The cloet work to our i that of Kim et al. [12]; they conider cheduling real-time tak on cloud ytem. However, in their work the real-time contraint i pecified in a ervice level agreement (SLA). In uch cloud model, the uer doe not elect individual VM and VM allocation i left to the ervice provider. Their work examine power management while allocating VM to meet the SLA. In our work, we conider the IaaS model where the uer elect and pay the cot for the VM (imilar to Amazon model), and we propoe a cheme to reduce cot for the uer. Thu the work of Kim et al. [12] benefit the ervice provider and our work benefit the uer. Thi paper ha the following contribution: (1) Thi i the firt paper to conider cot-efficient reource allocation in heterogenou cloud for real-time tak. (2) We formulate reource allocation a a contrained optimization problem. (3) We propoe a polynomial-time algorithm to allocate VM while meeting tak real-time contraint and we how how the cot varie baed on the ditribution of the tak et. III. REAL-TIME CLOUD SCHEDULING Thi ection decribe the problem of cheduling real-time tak. Section III-A and III-B define the problem, and how a imple example. Section III-C formulate a contrained optimization problem. Section III-D decribe a olution and Section III-E preent our algorithm to olve the problem. A. Problem Definition The application ha a et of tak T. Each tak α i T ha an arrival time a i and a deadline d i (pecified in minute). We ue cycle to quantify the tak workload w i ; the uage of cycle in thi context i a general meaure of the amount of computation required for the workload. The workload of each tak i parallelizable into maller unit (ubtak); the ize of the mallet poible ubtak i p. R i the et of available VM. Each VM v j R ha a computation peed j and the correponding cot c j. The peed j i the number of cycle the VM can complete per minute. The uer i charged the cot c j for renting v j for D minute continuouly, regardle of the utilization within the interval. D i the minimum time unit for renting. We aume that tak α i can alway meet it deadline d i if enough VM are allocated. The condition p i i, j, d i a i max( j) ; in other word, the fatet VM max( j ) can compute the mallet ubtak p before the deadline. The problem can be tated a follow: find an offline mapping from T onto a ubet of R to minimize the overall cot while meeting the deadline of all tak. B. Motivating Example We decribe a imple motivating example. Table III how the peed and the cot of the reource available for election. The cot are in the ame ratio a table I. The value of D i 60 minute and p i 10. We conider a ingle tak α 1 with an arrival time a 1 of 0 min, workload w 1 of 600 cycle,

3 (a) (b) (c) Fig. 1. Allocation of three reource v 1, v 2,andv 3 for two tak: α 1 i hown in olid rectangle, and α 2 in empty rectangle. (a) Auming v 1 i allocated to α 1 from 0 to 60 min, allocating two of v 2 at 40 min to α 2. The cot i =4.(b)Ifv 2 i allocated to α 1, v 2 can be alo ued for α 2 from 40 min to 60 min. Thi along with one of v 3 at 40 min can complete α 2. The cot i 1.5+2=3.5. (c)ifv 3 i allocated to α 1, v 3 can alo be ued for α 2 from 40 min to 80 min. Thi along with one of v 1 at 40 min can complete α 2 with the lowet total cot 2+1=3. and a deadline d 1 of 70 min. Tak α 1 take 60 (= w1 1 ), 30 (= w1 2 ), and 20 (= w1 3 ) minute on v 1, v 2, and v 3 repectively. The objective i to reduce the cot while meeting the deadline. A chedule elect the reource with the lowet cot while meeting α 1 deadline. Renting v 1 from 0 min to 60 min can finih α 1 before d 1 with the lowet cot. For each tak α i,weuec αi to denote the cot for the tak to meet it deadline and c α1 =1. VM v j j = cycle c min j = cot hour v v v TABLE III RESOURCES AVAILABLE FOR SELECTION Tak Arrival Deadline Workload α i a i d i w i α α TABLE IV TASKS TO BE COMPLETED Next, we conider tak α 2 with arrival time a 2 =40min, workload w 2 = 1500 cycle, and deadline d 2 =80min.We ue the election v 1 made earlier for α 1, and examine how to allocate reource to execute α 2. We need to complete α 2 in d 2 a 2 = 40 minute; thi can be accomplihed with four v 1 ( =150 minute, 40 =4v 1), two v 2 or, two v 3.The correponding cot for both tak are 1+4=5, 1+3=4, and 1+4=5 repectively. If we conider uing more than one type of VM for α 2, uing one v 3 and one v 2 reduce the cot to 1+(2+1.5) = 4.5. Uing one v 3 and one v 1 further reduce the cot to 1+(2+1) = 4 and thi i the lowet total cot. Figure 1(a) how the example with c α1 +c α2 =4, uing v 1 for α 1 and two of v 2 for α 2. If we conider both tak imultaneouly, a better olution allocate v 2 to α 1, and v 3 to α 2, hown in Figure 1(b). Thi reult in c α1 + c α2 = = 3.5, while meeting both deadline. The cot i lower becaue α 1 utilize v 2 from 0 minto30min,andα 2 can be run on v 2 from 40 min to 60 min, thu completing 400 cycle of α 2.Onev 3 can be rented at 40 min, and the remaining 1100 cycle of α 2 can be run on v 3 before the deadline d 2. The example how that a trategy may not give the lowet total cot. The tak mut be conidered imultaneouly to minimize the total cot. In the next ection, we formulate reource allocation a a contrained optimization problem. C. Problem Formulation We formulate the problem decribed in Section III-A a a contrained optimization problem. Each type of VM v j R ha peed j and cot c j, j =1,2,...x. We may elect any number of VM of a given type and we ue indicator variable δ(j, k) to repreent the election of VM. More pecifically, δ(j, k) = 1 if the k VM of type v j are ued, where k = 1, 2, 3... For example, δ(2, 3) = 1 indicate that three v 2 are ued. The objective function i to minimize the total cot. Thi may be given by umming the cot of all the elected VM a hown in equation (1): x y min c j δ(j, k) (1) j=1 k=1 The outer loop in equation (1) correpond to the x different type of VM, and the inner loop ummed to y correpond to the number of available VM of each type. Since cloud computing offer the election of an arbitrary number of VM, the value of y =. In reality, we can derive an upper bound for y for a given et of tak T. The total computation for T i given by C = T i=1 w i. The computation C i partitioned among the VM. If we conider the extreme cae that each VM perform the mallet unit p, then the maximum number of VM that can be ued for T i C p.the( C p +1)th VM cannot be ued a each of the previou C p VM i already performing the mallet unit of computation; there i no more computation for any additional VM. Thu we have an upper bound for y to be y C T p = 1 p w i. (2) i=1 Equation (1) repreent the total cot to be minimized and i the objective function. We now examine the contraint: all tak need to meet their deadline. We need to elect ufficient VM in equation (1) uch that each tak α i can be completed after it arrival time a i and before it deadline d i.theδ in equation (1) only indicate the type and the number of VM allocated; they do not indicate when the VM are allocated. The time of allocation i important to enure that the VM allocated for α i are available after a i and before d i. In order to conider the time of allocation, we make a retriction (for now) that each allocated VM i available for 1 minute (D=1), and we introduce a new et of indicator variable θ that include

4 the temporal information: θ(j, k, m) = 1, ifthek VMoftype j are available at the m th minute. Reconider the example in Section III-B. For tak α 1, a 1 = 0, d 1 =70, w 1 = 600, and p =10. To enure thi tak i completed before it deadline, VM mut be allocated uch that the number of cycle completed in the interval between a 1 (0 min) and d 1 (70 min) i at leat w 1 (600 cycle). The number of cycle completed at each minute in thi interval depend on the peed of the VM allocated at each minute within the interval; formulating thi uing θ give x y j θ(j, k, m) (3) j=1 k=1 cycle at the m th minute. In thi example, x i 3 becaue there are three different type of VM. From equation (2), the value of y cannot exceed =60. Thi mean that the maximum number of VM we can ue i 60, with each computing 10 cycle. In equation (3), if θ(2, 1, 1) = 1 and θ(1,k,1) = θ(3,k,1) = 0 for 1 k 60, the peed at the 1 t minute i 2 =20cycle/min. The amount of work that can be done during thi minute i 20 cycle. To enure that at leat 600 cycle are completed for the tak, we need to make ure the ummation of work done at each minute from 0 to 70 i at leat 600. To generalize thi acro the duration of tak α i T, we obtain the following requirement, for each α i : d i x y j θ(j, k, m) w i (4) m=a i j=1 k=l i The lower limit of k for the i th tak equal l i becaue ome θ are already allocated for previou i-1 tak (up to tak α i 1 ); thi enure that θ for the current tak α i doe not with the θ allocated for tak α 0, α 1... α i 1. Thi prevent haring VM among tak; however, our aumption that D = 1 enure that each VM i already fully utilized at the minute it i available. Equation (4) i baed on the aumption that D=1. In order to generalize our formulation uch that a VM i available for D minute, we need to define a function to repreent a time interval of D minute. We ue the window function u(m) u(m D), where u(m) i the unit tep function. We now need to find the number of D minute window to perform the ame amount of computation a the allocation for θ in equation (4). In order to do thi, we define equation (5) and (6) for each VM type j. At each time intant m, θ(j, k, m) hould be le than or equal to n (m z)j ; n (m z)j i the mallet number of D minute window required to do theameworkaθ(j, k, m) at time m. Weueaubcript of m z becaue a D-minute window that begin z minute before m, can be ued at m, a long a z D. Thevalueof m i between the arrival of the firt tak a 1 and the deadline of the lat tak d N. f(m, z) =u(m z) u(m z + D) d N y D ( θ(j, k, m) n (m z)j f(m, z)) 0 m=a 1 k=1 z=0 To find the total number of VM of each type j that are required, we define t j = d N m=a 1 D (5) n mj (6) Speed Cot VM ued v v v v 1 + v v 2 + v v 3 + v v 1 +2 v v 2 +2 v v 3 +2 v 3... TABLE V LOOK-UP TABLE FOR DIFFERENT SPEED REQUIREMENTS AND THE CORRESPONDING COSTS AND VMS. Equation (6) um all the n mj at every poible m, ranging from D minute before the arrival of the firt tak, to deadline of the lat tak. Thi um t j give the total number of VM of type j that are required. In order to make thi election a minimum cot election in equation (1), we et δ(j, k) =1 k t j. (7) The olution to thi ILP formulation i intractable. In Section III-D and III-E, we decribe a algorithm and our polynomial time heuritic to olve thi problem. D. EDF-Greedy Algorithm The firt olution in Section III-B can be decribed a a trategy baed on EDF (earliet deadline firt). The tak are conidered in the order of their deadline. The trategy firt trie to allocate a tak to VM available from the allocation for previou tak. If thee VM are inufficient to complete the tak before it deadline, the trategy elect the cheapet et of VM that can complete the tak before the deadline. To find the cheapet et of VM for each tak, we contruct a lookup table baed on the peed and cot in Table III. The lookup table contain a range of poible computing peed contructed from the different type of VM. In Table III, the VM have peed {10, 20, 30}. Uing thee peed a the bae, we can obtain the combinatorial et of peed S = {10, 20, 30, 40, 50, 60,...}. For each peed in S, thevm et to give that peed i identified and tored with that peed in Table V. For example, to achieve peed of 50, we can ue one v 2 and one v 3 with cot = =3.5, ortwov 2 and one v 1 with cot = =4. The former ha a lower cot and it i a better allocation. The entrie in the lookup table are orted in order of their peed. For a given tak workload, the trategy earche the lookup table to find the lowet peed that can finih the workload before the deadline. For tak α 1 in Section III-B, the lowet peed needed to finih the tak of w 1 = 600 cycle in d 1 a 1 =70minute i 10 cycle/min. If d 1 i changed to 10 min, then the lowet peed needed to compute 600 cycle in 10 min in the table i 60 cycle/min. If thi lowet peed in the lookup table i αi, the correponding cot i c αi.the lookup table i finite becaue the number of row i bounded by the highet peed needed to atify any tak. Thi bound i computed by w i max( ), α i T. (8) d i a i Thi guarantee that the trategy can find a VM allocation for all the tak. The complexity of computing the

5 lookup table depend on the number of type of VM x. In the above example, x =3and it i oberved that the VM have peed-cot ratio of 10, 13.33, and 15 for v 1, v 2, and v 3 repectively. We contruct the lookup table by trying to allocate VM in decreaing order of their peed-cot ratio, i.e., we try to allocate a many of v 3 before we allocate v 2, and o on. To get a peed x, we can compute the required VM to be x = t 2 of v 2, and 3 = t 3 of v 3, x (t3 3) 2 x (t3 3+t2 2) 1 of v 1. Thu a peed of 80 i obtained by uing =2ofv 3 (t 3 =2), and =1ofv 2, and 80 ( ) 10 =0ofv 1,givingv 2 +2 v 3, a een in Table V. Similar formulation may be obtained for different type of VM with different peed and cot. E. Allocation Conidering Temporal Overlap The optimization problem formulated in Section III-C may be olved by exhautive combination of allocating different type of reource to all tak at variou time intant. Thi approach unfortunately i intractible. The algorithm in Section III-D allocate VM for each tak eparately. When the tak in time, thi trategy may produce under-utilized and higher-cot allocation. Thi ection decribe our polynomial-time algorithm that conider all tak together in the order of their deadline. For each tak α i, the algorithm (1) identifie ping tak in the future and (2) allocate reource conidering thee tak. (1) Identifying Overlapping Tak: For a given tak α i, we conider the tak in the future whoe deadline are after d i and examine whether thee tak temporally with α i.takα j temporally with α i if a j <d i + D. Thi mean that a VM can be allocated uch that it i hared by α i and α j.weuet i to denote the et of tak that with α i : T i {α j d j d i, and a j <d i + D} (9) (2) Allocating Reource: Firt, we ue Table V to obtain the lowet peed αi required to compute tak α i (Section III-D). Next, we examine if allocating a VM with peed greater than αi for tak α i can benefit the ping tak in T i. We conider each VM peed αi in the lookup table and compute the VM time t ij for each tak α j T i. We define t ij a the amount of time the VM allocated for α i i till available for running α j.thevalueoft ij depend on the arrival time, workload, and deadline of α i and α j, and the peed of the VM. It i calculated by: min(d D ij,d j a j ) if a j a i wi t ij = D D ij, if a j a i < wi,d j a i D d j (a i + D ij ), if a j a i < wi,d j a i <D where D ij = w i + max(a j d i, 0). (10) In the above equation, the time to complete the α i i wi min. The firt cae how the condition when the arrival time of the two tak are ufficiently apart uch that α i i completed before the arrival time of α j, i.e. (a j a i wi ). Tak α j can ue the VM for t j min. The value of t j i given by the leer of two term, D D ij, and d j a j. The firt term D D ij denote remaining time for the VM for α j,after taking into account the time it i ued by α i,= wi, and a poible gap a j d i between the deadline of the firt tak and the arrival of the econd tak; during thi time, the VM cannot be ued by both tak. For example, aume the firt tak ue w 1 =10min of the VM, mut be completed by d 1 =20min, and the econd tak arrive at time a 2 =40min, and mut be completed by d 2 =60min. Then the value of D 12 from equation (10) i 10 + max((40 20), 0)=30, and D D 12 i 60 30=30. However, d 2 a 2 =60 40=20. Thu, the VM that i available for 30 min at a 2 i ued only for min(30,20) minute, and thi i t 12 for the firt cae in the above equation. The other two cae are imilarly contructed. Note that the VM may not be ued for the entire t ij if tak α j doe not have ufficient workload w j to utilize the VM for t ij min. In uch a cenario, the temporal by α j i reduced from t ij to wj where i the peed of the VM being conidered, αi.weuetheo to repreent the actual. O min( w j,t ij), for αi (11) The maller of wj and t ij i the amount of time a VM allocated for a previou tak can be actually ued for a future tak. Baed on the O, we compute a revied cot for the current tak α i. The revied cot c α i i not the actual cot paid for the reource, and i ued only for the purpoe of deciion making in our algorithm. The cot c α i i calculated c α i min{c αi (1 O D )}, α j T i, αi. (12) In the above equation, the cot c α i i obtained by reducing c αi baed on how much the VM with tak in the future (1 O D ). Since we elect the reource correponding to c α i for each tak α i, thi encourage election of VM that can be ued by other tak in the future. Our algorithm retrict the et of peed earched to the ize of the lookup table. Further, our algorithm conider the et T i (all tak that with the current tak) while making an allocation for α i. A more exhautive analyi could further conider that each tak α j T i ha a correponding et of tak T j that i alo affected by the allocation for α i, and o on. Conidering thi would make our algorithm cloer to exhautive earch. To illutrate our algorithm, we conider the tak α 1 and α 2 from Section III-B and the value of D =60. Originally, we had aigned v 1 to α 1 ince it ha the lowet cot (c α1 =1), and finihe before the deadline. We now examine the cot for VM allocation to α 1 for the temporal- algorithm. The between α 1 and α 2 begin at 40 min. If we allocate v 1 for α 1, the cot remain the ame (t 12 =0,O =0 and c α 1 =c α1 =1) ince α 1 ue v 1 for the entire 60 min, and v 1 cannot be made available for the ping tak α 2. Allocating v 2 for α 1 earlier reulted in c α1 = c 2 =1.5. Ifwe allocate v 2 for α 1 in the temporal- cheme, allocating it from 10 min to 70 min reult in v 2 being available for α 2 from 40 min to 70 min (t 12 = 30). The temporal O = min(30, ) = 30. The cot for thi allocation i c α 1 = c 2 (1 O 30 D ) = 1.5 (1 60 ) = 0.75, which i le than the previou option. Allocating v 3 for α 1 earlier reulted in c α1 = c 3 =2.For the temporal- cheme, if we allocate v 3 from 20 min to 80 min, α 1 i completed from 20 min to 40 min, and v 3 i

6 available for α 2 from 40 min to 80 min. Thi reult in a cot of c α 1 = c 3 c 3 ( )=2 2 ( 60 )=0.67, thu making the elector chooe v 3 for α 1 over v 1 and v 2. Aigning v 3 to α 1 a hown in Figure 1(c) reult in 1200 cycle of α 2 being completed from 40 min to 80 min; allocating v 1 for α 2 at 40 min can complete the tak with c α2 =1. The actual cot for α 1 and α 2 uing v 1 and v 3 i c α1 + c α2 = 2+1=3, lower than the total cot obtained in Section III-B. IV. EVALUATION In thi ection, we decribe how we evaluate the different cheduling algorithm. Section IV-A decribe the evaluation etup. Section IV-B compare the cot obtained by the algorithm. Section IV-C and IV-D decribe our enitivity analyi, and Section IV-E compare the complexity. A. Setup We conider three algorithm for our evaluation: (1) EDF (called ) decribed in Section III-D, (2) temporal- (called ) in Section III-E, and (3) exhautive earch, conidering all poible combination to obtain a minimum-cot allocation. It ue neted loop of the following, (1) each tak, (2) each type of VM, (3) any number of reource of a given type, and (4) every poible temporal allocation of each VM. The input to the cheme are the peed and cot of the reource available, and the arrival time, workload, deadline of the tak. We ue the peed and cot hown in Table III for our evaluation. We chooe a Poion ditribution with parameter λ to model the arrival time of the tak ince it i a natural way to model a equence of event randomly paced in time. We conider up to 500 tak for our evaluation and λ ranging from 0.5 to 500. We ue randomly generated deadline of the tak between (d l,d u ) minute and different value of d l and d u, ranging from 10 to The workload for the tak are randomly generated within a range of (p, max(w)) cycle, where the lower limit i p=10 cycle. We ue a range of upper limit for max(w), from 1500 to A decribed in Section III-D, the value of q and d l determine the number of entrie in the lookup table. Accordingly, we contruct the lookup table for the reource. B. Cot Analyi We oberve the cot returned by the cheme. The exhautive earch conider all poible allocation of reource, and return the lowet cot. However, ince exhautive earch take everal hour, we compare the cot returned by exhautive earch for only up to ten tak. The and cheme perform within 200% of the reult from exhautive earch for upto ten tak. For a greater number of tak, we only compare the and the cheme Figure 2 (a) how the cot for the and cheme, for different number of tak ranging from 100 to 500, with λ =1.5, d l =10, and d u = The value in the figure are averaged over 100 trial. The figure how that the cheme perform conitently better than the cheme. For 500 tak, the cheme return an average cot of 546, 22% reduction when compared to the value of 698. Thi i becaue the cheme conider haring VM temporally for ping tak. C. Varying Poion parameter λ The value of λ determine the ditribution of the tak. A lower value reult in a more concentrated ditribution, with more tak with a common arrival time. Figure 2 (b) how the hitogram of the arrival time of a mall et of tak from our evaluation, for different value of λ. It i oberved that the et of tak with lower value of λ have more in their arrival time. Thi in the arrival time correpond to an between the duration of the tak. Thu lower value of λ correpond to greater between the tak. We vary λ from 0.5 to 500 and oberve the cot returned by the and cheme in Figure 2 (c). The number of tak i 100. The cot returned by the cheme doe not vary ignificantly with different λ becaue the cheme doe not conider the tak imultaneouly. For the cheme, the cot obtained increae for higher value of λ. For example, for λ = 0.5, the cot returned by the cheme i 27.2% le than the cheme; for λ = 500, the improvement i reduced to 7.2%. Thi i becaue a λ increae, there i le between the tak, and hence conidering the tak imultaneouly ha le benefit. An extreme cae would be a cenario when there i no between the tak. In thi cae, the cheme would return the optimal olution. D. Varying workload Increaing the workload of the tak reult in a proportionate increae in the cot function. We vary the limit of the random generator (p, q) and oberve the variation in the cot. Figure 3(a) how the cot with p =10, for different value of q. Figure 3(b) how the cot with q = 4500 and different value of p. It i oberved that the cot for both cheme increae with the workload. E. Complexity analyi Exhautive earch ha exponential complexity and i not a calable olution. The cheme ha the following tep (1) Contruct the lookup table. The number of entrie i max( wi ) and each entry i linear in the number of different type of VM. Thu the complexity of thi tep i given by O(x max( wi ) ). (2) Viit each tak once and earch the entrie of the lookup table to find the allocation. Since the table i orted, earching can be done in O( max( wi ) ) time. Since there are N tak, thi reult in a complexity of O(N max( wi ) ). The overall complexity i the um of the above tep (1) and (2). If N>>xand max( wi ) i mall, the complexity of the olution i linear in the number of tak, i.e., O(N). For the cheme, we have the ame tep (1) and (2). In addition, for each tak, we conider all the tak with deadline greater than the current tak and examine the haring of VM between tak. Thi reult in each tak allocation having a complexity of O(N 1); ince there are N uch tak allocation, the overall complexity i given by O(N (N 1)) = O(N 2 ). Figure 3(c) how the time taken by the and cheme for different number of tak. We do not how the exhautive earch in the figure becaue it i exponential in the number of tak, number of reource, and type of reource and take everal hour for a et of 20 tak on a 3 GHz ytem with 4 GB RAM. We fit the data point in Figure 3(c), and we verify that the and

7 Cot Cot Number of tak (a) (b) Value of λ (c) Fig. 2. (a) Cot obtained by the and cheme, for different number of tak (lower cot i better). The value of λ i 1.5. (b) Hitogram of arrival time of the tak for different value of λ. The point in the hitogram are interpolated to how the ditribution of the arrival time. (c) Cot obtained by the cheme and the cheme for different value of λ for 100 tak. A the value of λ increae, the amount of decreae and thu the the cot obtained by the cheme increae. Cot Upper bound on workload q (a) Cot Lower bound on workload p (b) Time (m) Number of tak (c) Fig. 3. (a) the cot of the and cheme with lower bound p =10, for different value of upper bound q. (b) the cot of the and cheme with upper bound q = 4500 and different value of lower bound p. (c) performance of the and cheme for different number of tak. cheme have linear and quadratic time complexitie repectively. V. CONCLUSION We analyze the problem of allocating reource for real-time tak uch that the cot i minimized and all the deadline are met. We formulate the problem a a contrained optimization problem. A an optimal olution ha exponential complexity, we propoe an EDF- cheme and a cheme conider temporal ping to allocate reource efficiently. Our future work include extending the analyi by conidering tardine of tak, relaxed or oft real-time contraint, and conidering bulk dicount pricing. REFERENCES [1] T. A. AlEnawy et al. Energy-Aware Tak Allocation for Rate Monotonic Scheduling. In IEEE Real Time and Embedded Technology and Application Sympoium, page , March [2] U. Ali et al. Video baed Parallel Face recognition uing Gabor filter on homogeneou ditributed ytem. In IEEE International Conference on Engineering of Intelligent Sytem, page 1 5, [3] F. Aymerich et al. A real time financial ytem baed on grid and cloud computing. In ACM ympoium on Applied Computing, page , [4] J. Cong et al. Energy Efficient Multiproceor Tak Scheduling under Input-dependent Variation. In Deign, Automation and Tet in Europe, page , [5] R. Datta et al. Image Retrieval: Idea, Influence, and Trend of the New Age. ACM Computing Survey, 40:1 60, April [6] E. Deelman et al. The cot of doing cience on the cloud: the montage example. In ACM/IEEE Conference on Supercomputing, page 1 12, [7] K. Funaoka et al. Energy-Efficient Optimal Real-Time Scheduling on Multiproceor. In IEEE International Sympoium on Object/Component/Service-Oriented Real-Time Ditributed Computing, page 23 30, [8] S. Funk et al. Energy Minimization Technique for Real-Time Scheduling on Multiproceor Platform. In Technical Report 01-30, Computer Science Department, Univerity of North Carolina-Chapel Hill, [9] M. Goraczko et al. Energy-optimal oftware partitioning in heterogeneou multiproceor embedded ytem. In Deign Automation Conference, page , [10] R. Huang et al. Automatic reource pecification generation for reource election. In ACM/IEEE Conference on Supercomputing, page 1 11, [11] O. Kao et al. Scheduling apect for image retrieval in cluter-baed image databae. In IEEE/ACM Sympoium on Cluter Computing and Grid, page , [12] K. H. Kim et al. Power-aware proviioning of cloud reource for realtime ervice. In International Workhop on Middleware for Grid, Cloud and e-science, page 1 6, [13] W. Lee. Energy-Saving DVFS Scheduling of Multiple Periodic Real- Time Tak on Multi-core Proceor. In IEEE/ACM International Sympoium on Ditributed Simulation and Real Time Application, page , [14] S. Liu et al. On-Line Scheduling of Real-Time Service for Cloud Computing. In World Congre on Service, page , [15] C. Natar et al. Real-time face recognition uing feature combination. In IEEE International Conference on Automatic Face and Geture Recognition, page , [16] T. N takpé et al. A comparion of cheduling approache for mixedparallel application on heterogeneou platform. In International Sympoium on Parallel and Ditributed Computing, page 35, [17] K. Pua et al. Real time repeated video equence identification. Computer Viion and Image Undertanding, 93(3): , [18] F. Suter. Scheduling Delta Critical Tak in mixed-parallel application on a national grid. In IEEE/ACM International Conference on Grid Computing, page 2 9, [19] W. Tai et al. Real-Time Service-Oriented Cloud Computing. In World Congre on Service, page , [20] P. Viola et al. Robut real-time object detection. International Journal of Computer Viion, 57(2): , [21] G. Wei et al. A game-theoretic method of fair reource allocation for cloud computing ervice. The Journal of Supercomputing, page 1 18, [22] C. Xian et al. Energy-aware cheduling for real-time multiproceor ytem with uncertain tak execution time. In Deign Automation Conference, page 669, [23] G. Zeng et al. Practical Energy-Aware Scheduling for Real-Time Multiproceor Sytem. In IEEE International Conference on Embedded and Real-Time Computing Sytem and Application, page , 2009.

DISTRIBUTED DATA PARALLEL TECHNIQUES FOR CONTENT-MATCHING INTRUSION DETECTION SYSTEMS. G. Chapman J. Cleese E. Idle

DISTRIBUTED DATA PARALLEL TECHNIQUES FOR CONTENT-MATCHING INTRUSION DETECTION SYSTEMS. G. Chapman J. Cleese E. Idle DISTRIBUTED DATA PARALLEL TECHNIQUES FOR CONTENT-MATCHING INTRUSION DETECTION SYSTEMS G. Chapman J. Cleee E. Idle ABSTRACT Content matching i a neceary component of any ignature-baed network Intruion Detection

More information

DISTRIBUTED DATA PARALLEL TECHNIQUES FOR CONTENT-MATCHING INTRUSION DETECTION SYSTEMS

DISTRIBUTED DATA PARALLEL TECHNIQUES FOR CONTENT-MATCHING INTRUSION DETECTION SYSTEMS DISTRIBUTED DATA PARALLEL TECHNIQUES FOR CONTENT-MATCHING INTRUSION DETECTION SYSTEMS Chritopher V. Kopek Department of Computer Science Wake Foret Univerity Winton-Salem, NC, 2709 Email: kopekcv@gmail.com

More information

A note on profit maximization and monotonicity for inbound call centers

A note on profit maximization and monotonicity for inbound call centers A note on profit maximization and monotonicity for inbound call center Ger Koole & Aue Pot Department of Mathematic, Vrije Univeriteit Amterdam, The Netherland 23rd December 2005 Abtract We conider an

More information

Queueing systems with scheduled arrivals, i.e., appointment systems, are typical for frontal service systems,

Queueing systems with scheduled arrivals, i.e., appointment systems, are typical for frontal service systems, MANAGEMENT SCIENCE Vol. 54, No. 3, March 28, pp. 565 572 in 25-199 ein 1526-551 8 543 565 inform doi 1.1287/mnc.17.82 28 INFORMS Scheduling Arrival to Queue: A Single-Server Model with No-Show INFORMS

More information

Two Dimensional FEM Simulation of Ultrasonic Wave Propagation in Isotropic Solid Media using COMSOL

Two Dimensional FEM Simulation of Ultrasonic Wave Propagation in Isotropic Solid Media using COMSOL Excerpt from the Proceeding of the COMSO Conference 0 India Two Dimenional FEM Simulation of Ultraonic Wave Propagation in Iotropic Solid Media uing COMSO Bikah Ghoe *, Krihnan Balaubramaniam *, C V Krihnamurthy

More information

INFORMATION Technology (IT) infrastructure management

INFORMATION Technology (IT) infrastructure management IEEE TRANSACTIONS ON CLOUD COMPUTING, VOL. 2, NO. 1, MAY 214 1 Buine-Driven Long-term Capacity Planning for SaaS Application David Candeia, Ricardo Araújo Santo and Raquel Lope Abtract Capacity Planning

More information

A Spam Message Filtering Method: focus on run time

A Spam Message Filtering Method: focus on run time , pp.29-33 http://dx.doi.org/10.14257/atl.2014.76.08 A Spam Meage Filtering Method: focu on run time Sin-Eon Kim 1, Jung-Tae Jo 2, Sang-Hyun Choi 3 1 Department of Information Security Management 2 Department

More information

Cluster-Aware Cache for Network Attached Storage *

Cluster-Aware Cache for Network Attached Storage * Cluter-Aware Cache for Network Attached Storage * Bin Cai, Changheng Xie, and Qiang Cao National Storage Sytem Laboratory, Department of Computer Science, Huazhong Univerity of Science and Technology,

More information

Scheduling of Jobs and Maintenance Activities on Parallel Machines

Scheduling of Jobs and Maintenance Activities on Parallel Machines Scheduling of Job and Maintenance Activitie on Parallel Machine Chung-Yee Lee* Department of Indutrial Engineering Texa A&M Univerity College Station, TX 77843-3131 cylee@ac.tamu.edu Zhi-Long Chen** Department

More information

Optical Illusion. Sara Bolouki, Roger Grosse, Honglak Lee, Andrew Ng

Optical Illusion. Sara Bolouki, Roger Grosse, Honglak Lee, Andrew Ng Optical Illuion Sara Bolouki, Roger Groe, Honglak Lee, Andrew Ng. Introduction The goal of thi proect i to explain ome of the illuory phenomena uing pare coding and whitening model. Intead of the pare

More information

Group Mutual Exclusion Based on Priorities

Group Mutual Exclusion Based on Priorities Group Mutual Excluion Baed on Prioritie Karina M. Cenci Laboratorio de Invetigación en Sitema Ditribuido Univeridad Nacional del Sur Bahía Blanca, Argentina kmc@c.un.edu.ar and Jorge R. Ardenghi Laboratorio

More information

Bi-Objective Optimization for the Clinical Trial Supply Chain Management

Bi-Objective Optimization for the Clinical Trial Supply Chain Management Ian David Lockhart Bogle and Michael Fairweather (Editor), Proceeding of the 22nd European Sympoium on Computer Aided Proce Engineering, 17-20 June 2012, London. 2012 Elevier B.V. All right reerved. Bi-Objective

More information

Name: SID: Instructions

Name: SID: Instructions CS168 Fall 2014 Homework 1 Aigned: Wedneday, 10 September 2014 Due: Monday, 22 September 2014 Name: SID: Dicuion Section (Day/Time): Intruction - Submit thi homework uing Pandagrader/GradeScope(http://www.gradecope.com/

More information

Assessing the Discriminatory Power of Credit Scores

Assessing the Discriminatory Power of Credit Scores Aeing the Dicriminatory Power of Credit Score Holger Kraft 1, Gerald Kroiandt 1, Marlene Müller 1,2 1 Fraunhofer Intitut für Techno- und Wirtchaftmathematik (ITWM) Gottlieb-Daimler-Str. 49, 67663 Kaierlautern,

More information

Control of Wireless Networks with Flow Level Dynamics under Constant Time Scheduling

Control of Wireless Networks with Flow Level Dynamics under Constant Time Scheduling Control of Wirele Network with Flow Level Dynamic under Contant Time Scheduling Long Le and Ravi R. Mazumdar Department of Electrical and Computer Engineering Univerity of Waterloo,Waterloo, ON, Canada

More information

Mixed Method of Model Reduction for Uncertain Systems

Mixed Method of Model Reduction for Uncertain Systems SERBIAN JOURNAL OF ELECTRICAL ENGINEERING Vol 4 No June Mixed Method of Model Reduction for Uncertain Sytem N Selvaganean Abtract: A mixed method for reducing a higher order uncertain ytem to a table reduced

More information

CHARACTERISTICS OF WAITING LINE MODELS THE INDICATORS OF THE CUSTOMER FLOW MANAGEMENT SYSTEMS EFFICIENCY

CHARACTERISTICS OF WAITING LINE MODELS THE INDICATORS OF THE CUSTOMER FLOW MANAGEMENT SYSTEMS EFFICIENCY Annale Univeritati Apuleni Serie Oeconomica, 2(2), 200 CHARACTERISTICS OF WAITING LINE MODELS THE INDICATORS OF THE CUSTOMER FLOW MANAGEMENT SYSTEMS EFFICIENCY Sidonia Otilia Cernea Mihaela Jaradat 2 Mohammad

More information

Algorithms for Advance Bandwidth Reservation in Media Production Networks

Algorithms for Advance Bandwidth Reservation in Media Production Networks Algorithm for Advance Bandwidth Reervation in Media Production Network Maryam Barhan 1, Hendrik Moen 1, Jeroen Famaey 2, Filip De Turck 1 1 Department of Information Technology, Ghent Univerity imind Gaton

More information

NETWORK TRAFFIC ENGINEERING WITH VARIED LEVELS OF PROTECTION IN THE NEXT GENERATION INTERNET

NETWORK TRAFFIC ENGINEERING WITH VARIED LEVELS OF PROTECTION IN THE NEXT GENERATION INTERNET Chapter 1 NETWORK TRAFFIC ENGINEERING WITH VARIED LEVELS OF PROTECTION IN THE NEXT GENERATION INTERNET S. Srivatava Univerity of Miouri Kana City, USA hekhar@conrel.ice.umkc.edu S. R. Thirumalaetty now

More information

Apigee Edge: Apigee Cloud vs. Private Cloud. Evaluating deployment models for API management

Apigee Edge: Apigee Cloud vs. Private Cloud. Evaluating deployment models for API management Apigee Edge: Apigee Cloud v. Private Cloud Evaluating deployment model for API management Table of Content Introduction 1 Time to ucce 2 Total cot of ownerhip 2 Performance 3 Security 4 Data privacy 4

More information

SCM- integration: organiational, managerial and technological iue M. Caridi 1 and A. Sianei 2 Dipartimento di Economia e Produzione, Politecnico di Milano, Italy E-mail: maria.caridi@polimi.it Itituto

More information

A Note on Profit Maximization and Monotonicity for Inbound Call Centers

A Note on Profit Maximization and Monotonicity for Inbound Call Centers OPERATIONS RESEARCH Vol. 59, No. 5, September October 2011, pp. 1304 1308 in 0030-364X ein 1526-5463 11 5905 1304 http://dx.doi.org/10.1287/opre.1110.0990 2011 INFORMS TECHNICAL NOTE INFORMS hold copyright

More information

Partial optimal labeling search for a NP-hard subclass of (max,+) problems

Partial optimal labeling search for a NP-hard subclass of (max,+) problems Partial optimal labeling earch for a NP-hard ubcla of (max,+) problem Ivan Kovtun International Reearch and Training Center of Information Technologie and Sytem, Kiev, Uraine, ovtun@image.iev.ua Dreden

More information

RO-BURST: A Robust Virtualization Cost Model for Workload Consolidation over Clouds

RO-BURST: A Robust Virtualization Cost Model for Workload Consolidation over Clouds !111! 111!ttthhh IIIEEEEEEEEE///AAACCCMMM IIInnnttteeerrrnnnaaatttiiiooonnnaaalll SSSyyymmmpppoooiiiuuummm ooonnn CCCllluuuttteeerrr,,, CCClllooouuuddd aaannnddd GGGrrriiiddd CCCooommmpppuuutttiiinnnggg

More information

How Enterprises Can Build Integrated Digital Marketing Experiences Using Drupal

How Enterprises Can Build Integrated Digital Marketing Experiences Using Drupal How Enterprie Can Build Integrated Digital Marketing Experience Uing Drupal acquia.com 888.922.7842 1.781.238.8600 25 Corporate Drive, Burlington, MA 01803 How Enterprie Can Build Integrated Digital Marketing

More information

A Resolution Approach to a Hierarchical Multiobjective Routing Model for MPLS Networks

A Resolution Approach to a Hierarchical Multiobjective Routing Model for MPLS Networks A Reolution Approach to a Hierarchical Multiobjective Routing Model for MPLS Networ Joé Craveirinha a,c, Rita Girão-Silva a,c, João Clímaco b,c, Lúcia Martin a,c a b c DEEC-FCTUC FEUC INESC-Coimbra International

More information

TRADING rules are widely used in financial market as

TRADING rules are widely used in financial market as Complex Stock Trading Strategy Baed on Particle Swarm Optimization Fei Wang, Philip L.H. Yu and David W. Cheung Abtract Trading rule have been utilized in the tock market to make profit for more than a

More information

A Review On Software Testing In SDlC And Testing Tools

A Review On Software Testing In SDlC And Testing Tools www.ijec.in International Journal Of Engineering And Computer Science ISSN:2319-7242 Volume - 3 Iue -9 September, 2014 Page No. 8188-8197 A Review On Software Teting In SDlC And Teting Tool T.Amruthavalli*,

More information

Profitability of Loyalty Programs in the Presence of Uncertainty in Customers Valuations

Profitability of Loyalty Programs in the Presence of Uncertainty in Customers Valuations Proceeding of the 0 Indutrial Engineering Reearch Conference T. Doolen and E. Van Aken, ed. Profitability of Loyalty Program in the Preence of Uncertainty in Cutomer Valuation Amir Gandomi and Saeed Zolfaghari

More information

DUE to the small size and low cost of a sensor node, a

DUE to the small size and low cost of a sensor node, a 1992 IEEE TRANSACTIONS ON MOBILE COMPUTING, VOL. 14, NO. 10, OCTOBER 2015 A Networ Coding Baed Energy Efficient Data Bacup in Survivability-Heterogeneou Senor Networ Jie Tian, Tan Yan, and Guiling Wang

More information

A technical guide to 2014 key stage 2 to key stage 4 value added measures

A technical guide to 2014 key stage 2 to key stage 4 value added measures A technical guide to 2014 key tage 2 to key tage 4 value added meaure CONTENTS Introduction: PAGE NO. What i value added? 2 Change to value added methodology in 2014 4 Interpretation: Interpreting chool

More information

1 Introduction. Reza Shokri* Privacy Games: Optimal User-Centric Data Obfuscation

1 Introduction. Reza Shokri* Privacy Games: Optimal User-Centric Data Obfuscation Proceeding on Privacy Enhancing Technologie 2015; 2015 (2):1 17 Reza Shokri* Privacy Game: Optimal Uer-Centric Data Obfucation Abtract: Conider uer who hare their data (e.g., location) with an untruted

More information

Socially Optimal Pricing of Cloud Computing Resources

Socially Optimal Pricing of Cloud Computing Resources Socially Optimal Pricing of Cloud Computing Reource Ihai Menache Microoft Reearch New England Cambridge, MA 02142 t-imena@microoft.com Auman Ozdaglar Laboratory for Information and Deciion Sytem Maachuett

More information

Utility-Based Flow Control for Sequential Imagery over Wireless Networks

Utility-Based Flow Control for Sequential Imagery over Wireless Networks Utility-Baed Flow Control for Sequential Imagery over Wirele Networ Tomer Kihoni, Sara Callaway, and Mar Byer Abtract Wirele enor networ provide a unique et of characteritic that mae them uitable for building

More information

Mobile Network Configuration for Large-scale Multimedia Delivery on a Single WLAN

Mobile Network Configuration for Large-scale Multimedia Delivery on a Single WLAN Mobile Network Configuration for Large-cale Multimedia Delivery on a Single WLAN Huigwang Je, Dongwoo Kwon, Hyeonwoo Kim, and Hongtaek Ju Dept. of Computer Engineering Keimyung Univerity Daegu, Republic

More information

Proceedings of Power Tech 2007, July 1-5, Lausanne

Proceedings of Power Tech 2007, July 1-5, Lausanne Second Order Stochatic Dominance Portfolio Optimization for an Electric Energy Company M.-P. Cheong, Student Member, IEEE, G. B. Sheble, Fellow, IEEE, D. Berleant, Senior Member, IEEE and C.-C. Teoh, Student

More information

CASE STUDY BRIDGE. www.future-processing.com

CASE STUDY BRIDGE. www.future-processing.com CASE STUDY BRIDGE TABLE OF CONTENTS #1 ABOUT THE CLIENT 3 #2 ABOUT THE PROJECT 4 #3 OUR ROLE 5 #4 RESULT OF OUR COLLABORATION 6-7 #5 THE BUSINESS PROBLEM THAT WE SOLVED 8 #6 CHALLENGES 9 #7 VISUAL IDENTIFICATION

More information

Acceleration-Displacement Crash Pulse Optimisation A New Methodology to Optimise Vehicle Response for Multiple Impact Speeds

Acceleration-Displacement Crash Pulse Optimisation A New Methodology to Optimise Vehicle Response for Multiple Impact Speeds Acceleration-Diplacement Crah Pule Optimiation A New Methodology to Optimie Vehicle Repone for Multiple Impact Speed D. Gildfind 1 and D. Ree 2 1 RMIT Univerity, Department of Aeropace Engineering 2 Holden

More information

Auction-Based Resource Allocation for Sharing Cloudlets in Mobile Cloud Computing

Auction-Based Resource Allocation for Sharing Cloudlets in Mobile Cloud Computing 1 Auction-Baed Reource Allocation for Sharing Cloudlet in Mobile Cloud Computing A-Long Jin, Wei Song, Senior Member, IEEE, and Weihua Zhuang, Fellow, IEEE Abtract Driven by pervaive mobile device and

More information

Control Theory based Approach for the Improvement of Integrated Business Process Interoperability

Control Theory based Approach for the Improvement of Integrated Business Process Interoperability www.ijcsi.org 201 Control Theory baed Approach for the Improvement of Integrated Buine Proce Interoperability Abderrahim Taoudi 1, Bouchaib Bounabat 2 and Badr Elmir 3 1 Al-Qualadi Reearch & Development

More information

A New Optimum Jitter Protection for Conversational VoIP

A New Optimum Jitter Protection for Conversational VoIP Proc. Int. Conf. Wirele Commun., Signal Proceing (Nanjing, China), 5 pp., Nov. 2009 A New Optimum Jitter Protection for Converational VoIP Qipeng Gong, Peter Kabal Electrical & Computer Engineering, McGill

More information

Performance of Multiple TFRC in Heterogeneous Wireless Networks

Performance of Multiple TFRC in Heterogeneous Wireless Networks Performance of Multiple TFRC in Heterogeneou Wirele Network 1 Hyeon-Jin Jeong, 2 Seong-Sik Choi 1, Firt Author Computer Engineering Department, Incheon National Univerity, oaihjj@incheon.ac.kr *2,Correponding

More information

MANAGING DATA REPLICATION IN MOBILE AD- HOC NETWORK DATABASES (Invited Paper) *

MANAGING DATA REPLICATION IN MOBILE AD- HOC NETWORK DATABASES (Invited Paper) * MANAGING DATA REPLICATION IN MOBILE AD- HOC NETWORK DATABASES (Invited Paper) * Praanna Padmanabhan School of Computer Science The Univerity of Oklahoma Norman OK, USA praannap@yahoo-inc.com Dr. Le Gruenwald

More information

Online story scheduling in web advertising

Online story scheduling in web advertising Online tory cheduling in web advertiing Anirban Dagupta Arpita Ghoh Hamid Nazerzadeh Prabhakar Raghavan Abtract We tudy an online job cheduling problem motivated by toryboarding in web advertiing, where

More information

Performance of a Browser-Based JavaScript Bandwidth Test

Performance of a Browser-Based JavaScript Bandwidth Test Performance of a Brower-Baed JavaScript Bandwidth Tet David A. Cohen II May 7, 2013 CP SC 491/H495 Abtract An exiting brower-baed bandwidth tet written in JavaScript wa modified for the purpoe of further

More information

Return on Investment and Effort Expenditure in the Software Development Environment

Return on Investment and Effort Expenditure in the Software Development Environment International Journal of Applied Information ytem (IJAI) IN : 2249-0868 Return on Invetment and Effort Expenditure in the oftware Development Environment Dineh Kumar aini Faculty of Computing and IT, ohar

More information

Assigning Tasks for Efficiency in Hadoop

Assigning Tasks for Efficiency in Hadoop Aigning Tak for Efficiency in Hadoop [Extended Abtract] Michael J. Ficher Computer Science Yale Univerity P.O. Box 208285 New Haven, CT, USA michael.ficher@yale.edu Xueyuan Su Computer Science Yale Univerity

More information

Risk Management for a Global Supply Chain Planning under Uncertainty: Models and Algorithms

Risk Management for a Global Supply Chain Planning under Uncertainty: Models and Algorithms Rik Management for a Global Supply Chain Planning under Uncertainty: Model and Algorithm Fengqi You 1, John M. Waick 2, Ignacio E. Gromann 1* 1 Dept. of Chemical Engineering, Carnegie Mellon Univerity,

More information

License & SW Asset Management at CES Design Services

License & SW Asset Management at CES Design Services Licene & SW Aet Management at CES Deign Service johann.poechl@iemen.com www.ces-deignservice.com 2003 Siemen AG Öterreich Overview 1. Introduction CES Deign Service 2. Objective and Motivation 3. What

More information

REDUCTION OF TOTAL SUPPLY CHAIN CYCLE TIME IN INTERNAL BUSINESS PROCESS OF REAMER USING DOE AND TAGUCHI METHODOLOGY. Abstract. 1.

REDUCTION OF TOTAL SUPPLY CHAIN CYCLE TIME IN INTERNAL BUSINESS PROCESS OF REAMER USING DOE AND TAGUCHI METHODOLOGY. Abstract. 1. International Journal of Advanced Technology & Engineering Reearch (IJATER) REDUCTION OF TOTAL SUPPLY CHAIN CYCLE TIME IN INTERNAL BUSINESS PROCESS OF REAMER USING DOE AND Abtract TAGUCHI METHODOLOGY Mr.

More information

Auction Mechanisms Toward Efficient Resource Sharing for Cloudlets in Mobile Cloud Computing

Auction Mechanisms Toward Efficient Resource Sharing for Cloudlets in Mobile Cloud Computing 1 Auction Mechanim Toward Efficient Reource Sharing for Cloudlet in Mobile Cloud Computing A-Long Jin, Wei Song, Ping Wang, Duit Niyato, and Peijian Ju Abtract Mobile cloud computing offer an appealing

More information

6. Friction, Experiment and Theory

6. Friction, Experiment and Theory 6. Friction, Experiment and Theory The lab thi wee invetigate the rictional orce and the phyical interpretation o the coeicient o riction. We will mae ue o the concept o the orce o gravity, the normal

More information

Growth and Sustainability of Managed Security Services Networks: An Economic Perspective

Growth and Sustainability of Managed Security Services Networks: An Economic Perspective Growth and Sutainability of Managed Security Service etwork: An Economic Perpective Alok Gupta Dmitry Zhdanov Department of Information and Deciion Science Univerity of Minneota Minneapoli, M 55455 (agupta,

More information

Towards Control-Relevant Forecasting in Supply Chain Management

Towards Control-Relevant Forecasting in Supply Chain Management 25 American Control Conference June 8-1, 25. Portland, OR, USA WeA7.1 Toward Control-Relevant Forecating in Supply Chain Management Jay D. Schwartz, Daniel E. Rivera 1, and Karl G. Kempf Control Sytem

More information

Laureate Network Products & Services Copyright 2013 Laureate Education, Inc.

Laureate Network Products & Services Copyright 2013 Laureate Education, Inc. Laureate Network Product & Service Copyright 2013 Laureate Education, Inc. KEY Coure Name Laureate Faculty Development...3 Laureate Englih Program...9 Language Laureate Signature Product...12 Length Laureate

More information

Health Insurance and Social Welfare. Run Liang. China Center for Economic Research, Peking University, Beijing 100871, China,

Health Insurance and Social Welfare. Run Liang. China Center for Economic Research, Peking University, Beijing 100871, China, Health Inurance and Social Welfare Run Liang China Center for Economic Reearch, Peking Univerity, Beijing 100871, China, Email: rliang@ccer.edu.cn and Hao Wang China Center for Economic Reearch, Peking

More information

Bidding for Representative Allocations for Display Advertising

Bidding for Representative Allocations for Display Advertising Bidding for Repreentative Allocation for Diplay Advertiing Arpita Ghoh, Preton McAfee, Kihore Papineni, and Sergei Vailvitkii Yahoo! Reearch. {arpita, mcafee, kpapi, ergei}@yahoo-inc.com Abtract. Diplay

More information

Tap Into Smartphone Demand: Mobile-izing Enterprise Websites by Using Flexible, Open Source Platforms

Tap Into Smartphone Demand: Mobile-izing Enterprise Websites by Using Flexible, Open Source Platforms Tap Into Smartphone Demand: Mobile-izing Enterprie Webite by Uing Flexible, Open Source Platform acquia.com 888.922.7842 1.781.238.8600 25 Corporate Drive, Burlington, MA 01803 Tap Into Smartphone Demand:

More information

T-test for dependent Samples. Difference Scores. The t Test for Dependent Samples. The t Test for Dependent Samples. s D

T-test for dependent Samples. Difference Scores. The t Test for Dependent Samples. The t Test for Dependent Samples. s D The t Tet for ependent Sample T-tet for dependent Sample (ak.a., Paired ample t-tet, Correlated Group eign, Within- Subject eign, Repeated Meaure,.. Repeated-Meaure eign When you have two et of core from

More information

Distributed Monitoring and Aggregation in Wireless Sensor Networks

Distributed Monitoring and Aggregation in Wireless Sensor Networks Ditributed Monitoring and Aggregation in Wirele Senor Network Changlei Liu and Guohong Cao Department of Computer Science & Engineering The Pennylvania State Univerity E-mail: {chaliu, gcao}@ce.pu.edu

More information

Review of Multiple Regression Richard Williams, University of Notre Dame, http://www3.nd.edu/~rwilliam/ Last revised January 13, 2015

Review of Multiple Regression Richard Williams, University of Notre Dame, http://www3.nd.edu/~rwilliam/ Last revised January 13, 2015 Review of Multiple Regreion Richard William, Univerity of Notre Dame, http://www3.nd.edu/~rwilliam/ Lat revied January 13, 015 Aumption about prior nowledge. Thi handout attempt to ummarize and yntheize

More information

Network Architecture for Joint Failure Recovery and Traffic Engineering

Network Architecture for Joint Failure Recovery and Traffic Engineering Network Architecture for Joint Failure Recovery and Traffic Engineering Martin Suchara Dept. of Computer Science Princeton Univerity, NJ 08544 muchara@princeton.edu Dahai Xu AT&T Lab Reearch Florham Park,

More information

Progress 8 measure in 2016, 2017, and 2018. Guide for maintained secondary schools, academies and free schools

Progress 8 measure in 2016, 2017, and 2018. Guide for maintained secondary schools, academies and free schools Progre 8 meaure in 2016, 2017, and 2018 Guide for maintained econdary chool, academie and free chool July 2016 Content Table of figure 4 Summary 5 A ummary of Attainment 8 and Progre 8 5 Expiry or review

More information

QUANTIFYING THE BULLWHIP EFFECT IN THE SUPPLY CHAIN OF SMALL-SIZED COMPANIES

QUANTIFYING THE BULLWHIP EFFECT IN THE SUPPLY CHAIN OF SMALL-SIZED COMPANIES Sixth LACCEI International Latin American and Caribbean Conference for Engineering and Technology (LACCEI 2008) Partnering to Succe: Engineering, Education, Reearch and Development June 4 June 6 2008,

More information

CASE STUDY ALLOCATE SOFTWARE

CASE STUDY ALLOCATE SOFTWARE CASE STUDY ALLOCATE SOFTWARE allocate caetud y TABLE OF CONTENTS #1 ABOUT THE CLIENT #2 OUR ROLE #3 EFFECTS OF OUR COOPERATION #4 BUSINESS PROBLEM THAT WE SOLVED #5 CHALLENGES #6 WORKING IN SCRUM #7 WHAT

More information

MECH 2110 - Statics & Dynamics

MECH 2110 - Statics & Dynamics Chapter D Problem 3 Solution 1/7/8 1:8 PM MECH 11 - Static & Dynamic Chapter D Problem 3 Solution Page 7, Engineering Mechanic - Dynamic, 4th Edition, Meriam and Kraige Given: Particle moving along a traight

More information

Comparison of Scheduling Algorithms for a Multi-Product Batch-Chemical Plant with a Generalized Serial Network

Comparison of Scheduling Algorithms for a Multi-Product Batch-Chemical Plant with a Generalized Serial Network Comparion of Scheduling Algorithm for a Multi-Product Batch-Chemical Plant with a Generalized Serial Network Niem-Trung L. Tra Thei ubmitted to the faculty of the Virginia Polytechnic Intitute and State

More information

Benchmarking Bottom-Up and Top-Down Strategies for SPARQL-to-SQL Query Translation

Benchmarking Bottom-Up and Top-Down Strategies for SPARQL-to-SQL Query Translation Benchmarking Bottom-Up and Top-Down Strategie for SPARQL-to-SQL Query Tranlation Kahlev a, Chebotko b,c, John Abraham b, Pearl Brazier b, and Shiyong Lu a a Department of Computer Science, Wayne State

More information

FEDERATION OF ARAB SCIENTIFIC RESEARCH COUNCILS

FEDERATION OF ARAB SCIENTIFIC RESEARCH COUNCILS Aignment Report RP/98-983/5/0./03 Etablihment of cientific and technological information ervice for economic and ocial development FOR INTERNAL UE NOT FOR GENERAL DITRIBUTION FEDERATION OF ARAB CIENTIFIC

More information

Morningstar Fixed Income Style Box TM Methodology

Morningstar Fixed Income Style Box TM Methodology Morningtar Fixed Income Style Box TM Methodology Morningtar Methodology Paper Augut 3, 00 00 Morningtar, Inc. All right reerved. The information in thi document i the property of Morningtar, Inc. Reproduction

More information

Project Management Basics

Project Management Basics Project Management Baic A Guide to undertanding the baic component of effective project management and the key to ucce 1 Content 1.0 Who hould read thi Guide... 3 1.1 Overview... 3 1.2 Project Management

More information

SELF-MANAGING PERFORMANCE IN APPLICATION SERVERS MODELLING AND DATA ARCHITECTURE

SELF-MANAGING PERFORMANCE IN APPLICATION SERVERS MODELLING AND DATA ARCHITECTURE SELF-MANAGING PERFORMANCE IN APPLICATION SERVERS MODELLING AND DATA ARCHITECTURE RAVI KUMAR G 1, C.MUTHUSAMY 2 & A.VINAYA BABU 3 1 HP Bangalore, Reearch Scholar JNTUH, Hyderabad, India, 2 Yahoo, Bangalore,

More information

Solutions to Sample Problems for Test 3

Solutions to Sample Problems for Test 3 22 Differential Equation Intructor: Petronela Radu November 8 25 Solution to Sample Problem for Tet 3 For each of the linear ytem below find an interval in which the general olution i defined (a) x = x

More information

CLUSTBIGFIM-FREQUENT ITEMSET MINING OF BIG DATA USING PRE-PROCESSING BASED ON MAPREDUCE FRAMEWORK

CLUSTBIGFIM-FREQUENT ITEMSET MINING OF BIG DATA USING PRE-PROCESSING BASED ON MAPREDUCE FRAMEWORK CLUSTBIGFIM-FREQUENT ITEMSET MINING OF BIG DATA USING PRE-PROCESSING BASED ON MAPREDUCE FRAMEWORK Sheela Gole 1 and Bharat Tidke 2 1 Department of Computer Engineering, Flora Intitute of Technology, Pune,

More information

How to Maximize User Satisfaction Degree in Multi-service IP Networks

How to Maximize User Satisfaction Degree in Multi-service IP Networks How to Maximize Uer Satifaction Degree in Multi-ervice IP Network Huy Anh Nguyen, Tam Van Nguyen and Deokai Choi Department of Electronic and Computer Engineering Chonnam National Univerity Gwangu, KOREA

More information

Module 8. Three-phase Induction Motor. Version 2 EE IIT, Kharagpur

Module 8. Three-phase Induction Motor. Version 2 EE IIT, Kharagpur Module 8 Three-phae Induction Motor Verion EE IIT, Kharagpur Leon 33 Different Type of Starter for Induction Motor (IM Verion EE IIT, Kharagpur Inructional Objective Need of uing arter for Induction motor

More information

AN OVERVIEW ON CLUSTERING METHODS

AN OVERVIEW ON CLUSTERING METHODS IOSR Journal Engineering AN OVERVIEW ON CLUSTERING METHODS T. Soni Madhulatha Aociate Preor, Alluri Intitute Management Science, Warangal. ABSTRACT Clutering i a common technique for tatitical data analyi,

More information

MSc Financial Economics: International Finance. Bubbles in the Foreign Exchange Market. Anne Sibert. Revised Spring 2013. Contents

MSc Financial Economics: International Finance. Bubbles in the Foreign Exchange Market. Anne Sibert. Revised Spring 2013. Contents MSc Financial Economic: International Finance Bubble in the Foreign Exchange Market Anne Sibert Revied Spring 203 Content Introduction................................................. 2 The Mone Market.............................................

More information

INTERACTIVE TOOL FOR ANALYSIS OF TIME-DELAY SYSTEMS WITH DEAD-TIME COMPENSATORS

INTERACTIVE TOOL FOR ANALYSIS OF TIME-DELAY SYSTEMS WITH DEAD-TIME COMPENSATORS INTERACTIVE TOOL FOR ANALYSIS OF TIMEDELAY SYSTEMS WITH DEADTIME COMPENSATORS Joé Lui Guzmán, Pedro García, Tore Hägglund, Sebatián Dormido, Pedro Alberto, Manuel Berenguel Dep. de Lenguaje y Computación,

More information

Queueing Models for Multiclass Call Centers with Real-Time Anticipated Delays

Queueing Models for Multiclass Call Centers with Real-Time Anticipated Delays Queueing Model for Multicla Call Center with Real-Time Anticipated Delay Oualid Jouini Yve Dallery Zeynep Akşin Ecole Centrale Pari Koç Univerity Laboratoire Génie Indutriel College of Adminitrative Science

More information

Optimizing a Semantic Comparator using CUDA-enabled Graphics Hardware

Optimizing a Semantic Comparator using CUDA-enabled Graphics Hardware Optimizing a Semantic Comparator uing CUDA-enabled Graphic Hardware Aalap Tripathy Suneil Mohan, Rabi Mahapatra Embedded Sytem and Codeign Lab codeign.ce.tamu.edu (Preented at ICSC 0, September 9, 0 in

More information

Growth and Sustainability of Managed Security Services Networks: An Economic Perspective

Growth and Sustainability of Managed Security Services Networks: An Economic Perspective Growth and Sutainability of Managed Security Service etwork: An Economic Perpective Alok Gupta Dmitry Zhdanov Department of Information and Deciion Science Univerity of Minneota Minneapoli, M 55455 (agupta,

More information

Mobility Improves Coverage of Sensor Networks

Mobility Improves Coverage of Sensor Networks Mobility Improve Coverage of Senor Networ Benyuan Liu Dept. of Computer Science Univerity of Maachuett-Lowell Lowell, MA 1854 Peter Bra Dept. of Computer Science City College of New Yor New Yor, NY 131

More information

v = x t = x 2 x 1 t 2 t 1 The average speed of the particle is absolute value of the average velocity and is given Distance travelled t

v = x t = x 2 x 1 t 2 t 1 The average speed of the particle is absolute value of the average velocity and is given Distance travelled t Chapter 2 Motion in One Dimenion 2.1 The Important Stuff 2.1.1 Poition, Time and Diplacement We begin our tudy of motion by conidering object which are very mall in comparion to the ize of their movement

More information

CHAPTER 5 BROADBAND CLASS-E AMPLIFIER

CHAPTER 5 BROADBAND CLASS-E AMPLIFIER CHAPTER 5 BROADBAND CLASS-E AMPLIFIER 5.0 Introduction Cla-E amplifier wa firt preented by Sokal in 1975. The application of cla- E amplifier were limited to the VHF band. At thi range of frequency, cla-e

More information

Despeckling Synthetic Aperture Radar Images with Cloud Computing using Graphics Processing Units

Despeckling Synthetic Aperture Radar Images with Cloud Computing using Graphics Processing Units Depeckling Synthetic Aperture Radar Image with Cloud Computing uing Graphic Proceing Unit Matej Keneman 1, Dušan Gleich, Amor Chowdhury 1 1 Margento R & D d.o.o., Gopovetka ceta 84, Maribor, Slovenia matej.keneman@gmail.com

More information

Sector Concentration in Loan Portfolios and Economic Capital. Abstract

Sector Concentration in Loan Portfolios and Economic Capital. Abstract Sector Concentration in Loan Portfolio and Economic Capital Klau Düllmann and Nancy Machelein 2 Thi verion: September 2006 Abtract The purpoe of thi paper i to meaure the potential impact of buine-ector

More information

Senior Thesis. Horse Play. Optimal Wagers and the Kelly Criterion. Author: Courtney Kempton. Supervisor: Professor Jim Morrow

Senior Thesis. Horse Play. Optimal Wagers and the Kelly Criterion. Author: Courtney Kempton. Supervisor: Professor Jim Morrow Senior Thei Hore Play Optimal Wager and the Kelly Criterion Author: Courtney Kempton Supervior: Profeor Jim Morrow June 7, 20 Introduction The fundamental problem in gambling i to find betting opportunitie

More information

Multi-Objective Optimization for Sponsored Search

Multi-Objective Optimization for Sponsored Search Multi-Objective Optimization for Sponored Search Yilei Wang 1,*, Bingzheng Wei 2, Jun Yan 2, Zheng Chen 2, Qiao Du 2,3 1 Yuanpei College Peking Univerity Beijing, China, 100871 (+86)15120078719 wangyileipku@gmail.com

More information

THE ECONOMIC INCENTIVES OF PROVIDING NETWORK SECURITY SERVICES ON THE INTERNET INFRASTRUCTURE

THE ECONOMIC INCENTIVES OF PROVIDING NETWORK SECURITY SERVICES ON THE INTERNET INFRASTRUCTURE THE ECONOMIC INCENTIVES OF PROVIDING NETWORK SECURITY SERVICES ON THE INTERNET INFRASTRUCTURE Li-Chiou Chen Department of Information Sytem School of Computer Science and Information Sytem Pace Univerity

More information

Cost Models for Selecting Materialized Views in Public Clouds

Cost Models for Selecting Materialized Views in Public Clouds International Journal of Data Warehouing and Mining, 0(4), -25, October-December 204 Cot Model for Selecting Materialized View in ublic Cloud Romain erriot, Clermont Univerité, Univerité Blaie acal, Aubière

More information

Imagery Portal Workshop #2 Department of Administrative Services, Executive Building Salem, Oregon May 11, 2006

Imagery Portal Workshop #2 Department of Administrative Services, Executive Building Salem, Oregon May 11, 2006 ry Portal Workhop #2 Department of Adminitrative Service, Executive Building Salem, Oregon May 11, 2006 Workhop Purpoe: dicu the outcome of the phae 1 coping proce for development of an imagery portal

More information

THE ECONOMIC INCENTIVES OF PROVIDING NETWORK SECURITY SERVICES ON THE INTERNET INFRASTRUCTURE

THE ECONOMIC INCENTIVES OF PROVIDING NETWORK SECURITY SERVICES ON THE INTERNET INFRASTRUCTURE Journal of Information Technology Management ISSN #1042-1319 A Publication of the Aociation of Management THE ECONOMIC INCENTIVES OF PROVIDING NETWORK SECURITY SERVICES ON THE INTERNET INFRASTRUCTURE LI-CHIOU

More information

OPINION PIECE. It s up to the customer to ensure security of the Cloud

OPINION PIECE. It s up to the customer to ensure security of the Cloud OPINION PIECE It up to the cutomer to enure ecurity of the Cloud Content Don t outource what you don t undertand 2 The check lit 2 Step toward control 4 Due Diligence 4 Contract 4 E-dicovery 4 Standard

More information

Computing Location from Ambient FM Radio Signals

Computing Location from Ambient FM Radio Signals Computing Location from Ambient FM Radio Signal Adel Youef Department of Computer Science Univerity of Maryland A.V. William Building College Park, MD 20742 adel@c.umd.edu John Krumm, Ed Miller, Gerry

More information

FOUR EASY PIECES FOR ASSESSING THE USABILITY OF MULTIMODAL INTERACTION: THE CARE PROPERTIES

FOUR EASY PIECES FOR ASSESSING THE USABILITY OF MULTIMODAL INTERACTION: THE CARE PROPERTIES 1 FOUR EASY PIECES FOR ASSESSING THE USABILITY OF MULTIMODAL INTERACTION: THE CARE PROPERTIES Daniel Salber BP 53, 38041 Grenoble Cedex 9, France {joelle.coutaz, laurence.nigay, daniel.alber} @imag.fr

More information

HUMAN CAPITAL AND THE FUTURE OF TRANSITION ECONOMIES * Michael Spagat Royal Holloway, University of London, CEPR and Davidson Institute.

HUMAN CAPITAL AND THE FUTURE OF TRANSITION ECONOMIES * Michael Spagat Royal Holloway, University of London, CEPR and Davidson Institute. HUMAN CAPITAL AND THE FUTURE OF TRANSITION ECONOMIES * By Michael Spagat Royal Holloway, Univerity of London, CEPR and Davidon Intitute Abtract Tranition economie have an initial condition of high human

More information

Parallel Adaptive Multigrid Methods in Plane Linear Elasticity Problems

Parallel Adaptive Multigrid Methods in Plane Linear Elasticity Problems Numerical Linear Algebra with Application, Vol. 1(1), 1 1 (1996) Parallel Adaptive Multigrid Method in Plane Linear Elaticity Problem Peter Batian, Stefan Lang Intitut für Computeranwendungen III, Univerität

More information

Managing Customer Arrivals in Service Systems with Multiple Servers

Managing Customer Arrivals in Service Systems with Multiple Servers Managing Cutomer Arrival in Service Sytem with Multiple Server Chrito Zacharia Department of Management Science, School of Buine Adminitration, Univerity of Miami, Coral Gable, FL 3346. czacharia@bu.miami.edu

More information

A Life Contingency Approach for Physical Assets: Create Volatility to Create Value

A Life Contingency Approach for Physical Assets: Create Volatility to Create Value A Life Contingency Approach for Phyical Aet: Create Volatility to Create Value homa Emil Wendling 2011 Enterprie Rik Management Sympoium Society of Actuarie March 14-16, 2011 Copyright 2011 by the Society

More information