Stochastic Analysis of Power-Aware Scheduling

Size: px
Start display at page:

Download "Stochastic Analysis of Power-Aware Scheduling"

Transcription

1 Stochatic Analyi of Power-Aware Scheduling Adam Wierman Computer Science Department California Intitute of Technology Lachlan L.H. Andrew Computer Science Department California Intitute of Technology Ao Tang School of ECE Cornell Univerity Abtract Energy conumption in a computer ytem can be reduced by dynamic peed caling, which adapt the proceing peed to the current load. Thi paper tudie the way to adjut peed to balance mean repone time and mean energy conumption, when job arrive a a Poion proce and proceor haring cheduling i ued. Both bound and aymptotic for the peed are provided. Interetingly, a imple cheme that halt when the ytem i idle and ue a tatic rate while the ytem i buy provide nearly the ame performance a the dynamic peed caling. However, dynamic peed caling which allocate a higher peed when more job are preent ignificantly improve robutne to burty traffic and mi-etimation of workload parameter. I. INTRODUCTION Two threat to the growth of the internet have their root in power conumption. The mot preing i that Moore law ha increaed the thermal denity of electronic to uch an extent that cooling i a major concern, and ha halted the previouly inexorable increae in clock peed. The longer term threat i that the need to reduce foil fuel conumption require all apect of ociety to conerve energy, while aggregate internet energy conumption i a ignificant and growing fraction of the energy conumption of developed countrie [1]. A a reult, all modern ytem deign mut conider the tradeoff between energy ue and other performance metric. Power can often be aved imply by running device more lowly. Dynamic peed caling, which electively reduce the peed when the load i light, reduce energy conumption with minimal impact on performance. It i widely implemented in current proceor, in the form of Intel SpeedStep and AMD PowerNow. Much of the theory of dynamic peed caling [2] [6] conider wort-cae bound. While uch wort-cae reult are important for temperature management [7], or when the energy to a pecific computer i contrained [8], [9], global energy conumption i affected by the average cae, rather than the wort cae. Conequently, thi paper tudie the average performance in a tochatic etting. In particular, thi paper eek to minimize a weighted um of the mean repone time and the energy ue per job. Thi performance metric ha been tudied both theoretically [1] [12] and in implementation [13]. Algorithm are known [12] for finding the peed which optimize thi objective in a tochatic M/M/1/FCFS etting. However, thee are highly recurive, and provide little inight. The goal of thi paper i to identify imple tructural propertie of the olution, and to ue them to compare the gain of uncontrained peed caling with that of an optimized tatic deign. The paper make three main contribution. Firt, the paper provide bound on the performance of dynamic peed caling (Section IV-A). Surpriingly, thee bound how that even an idealized verion of dynamic peed caling improve performance only marginally compared to a imple cheme where the erver ue a tatic peed when buy and run at peed when idle at mot a factor of 2 for typical parameter and often le (ee Section V). Second, the paper provide bound and aymptotic for the peed ued by the dynamic peed caling cheme (Section IV-B and IV-C). Thee reult provide inight into how the peed cale with the arriving load, the queue length, and the relative cot of energy. Third, the paper illutrate through analytic reult and numerical experiment that, though dynamic peed caling provide limited performance gain, it dramatically improve robutne to mi-etimation of workload parameter and burty traffic (Section VI). Note that many proof are omitted from thi document; all proof can be found in [14]. II. MODEL AND NOTATION In order to tudy the performance of dynamic peed caling, we focu on a imple model: an M/GI/1 PS queue with controllable ervice rate, dependent on the queue length. In thi model, job arrive to the erver a a Poion proce with rate λ, have intrinic ize with mean 1/µ, and depart at rate n µ when there are n job in the ytem. Under tatic cheme, the (contant) ervice rate i denoted by. Define the load a ρ = λ/µ, and note that the ρ i not the fraction of time the erver i buy. The performance metric we conider i E[T ] E[E]/β, where T i the repone time of a job, E i the energy expended on a job, and β repreent how delay-avere the deign i. It i often convenient to work with the expected cot per unit time, intead of per job, which by Little law can be written a z = E[N] λe[f()]/β, where N i the number of job in the ytem and f() determine the power ued when running at peed. The remaining piece of the model i to define the form of f(). The dynamic power of a circuit i typically a low-order polynomial in the peed [15]. A a reult, we will model the power ued by running at peed by λ f() β = α β where α > 1 and β take the role of β, but ha dimenion (1)

2 2 (time) α. The cot per unit time then become z = E[N] α β. (2) We will often focu on the cae of α = 2 to provide intuition. The impact of the workload parameter ρ, β, and α can often be captured by = ρ/β 1/α, which i a dimenionle meaure. Alo, it will often be convenient to ue the a natural dimenionle unit of peed /β 1/α. III. POWER-AWARE SPEED SELECTION Thi paper conider two natural form of peed caling: (i) Gated tatic peed: The erver gate it clock (etting = ) if no job are preent, and if job are preent it work at a contant rate choen to balance energy uage and repone time. (ii) Dynamic peed caling: The erver adapt it peed to the current number of requet preent in the ytem. The goal of thi paper i to undertand how to chooe peed in each of thee cenario and to contrat the relative merit of each cheme. Clearly the expected cot i reduced each time the erver i allowed to adjut it peed more dynamically. Thi mut be traded againt the cot of witching, uch a a delay of up to ten of microecond to change peed [16]. The important quetion i What i the magnitude of improvement at each level? For our comparion, we will ue idealized verion of each cheme. In particular, in each cae we will aume that the erver can be run at any deired peed in [, ) and ignore witching cot. In thi ection, we will derive expreion for the peed in cae (i). For cae (ii), we will decribe a numerical approach for calculating the peed which i due to George and Harrion [12]. Though thi numerical approach i efficient, it provide little inight into the tructure of the dynamic peed or the overall performance. Providing uch reult will be the focu of Section IV. A. The tatic peed for a gated ytem In the implet dynamic peed caling, a erver either run at a contant rate, or ha it clock gated uing zero dynamic power when the ytem i empty. We call thi policy the gated-tatic policy, and denote it cot z g. Since the erver can gate it clock, the energy cot i only incurred ρ/ of the time, when the erver i buy. Thu z = ρ ρα 1 ρ β. The optimum occur when > ρ and (α 1) α 2 ( ρ) 2 = β. (3) When α = 2, g = ρ β. In general, define G(; α) = σ.t. σ > (α 1)σ α (1 /σ) 2 = 1. (4) The gated-tatic peed i g = β 1/α G(; α). The following lemma bound G. Lemma 1. For α 2, 2 α α 1 G(; α) (α 1) 1/α 2 α (5) and the inequalitie are revered for α 2. B. Optimal dynamic peed caling A popular alternative to tatic power management i to allow the peed to adjut dynamically to the number of requet in the ytem. The tak of deigning an dynamic peed caling cheme in our model can be viewed a a tochatic control problem. We tart the analyi by noting that we can implify the problem dramatically with the following obervation. An M/GI/1 PS ytem i well-known to be inenitive to the job ize ditribution. Thi till hold when the ervice rate i queue-length dependent ince the policy till fall into the cla of ymmetric policie introduced by Kelly [17]. A a reult, the mean repone time and entire queue length ditribution are affected by the ervice ditribution through only it mean. Thu, we can conider an M/M/1 PS ytem. Further, the mean repone time and entire queue length ditribution are equivalent under all non-ize baed ervice ditribution in the M/M/1 queue [17]. Thu, to determine the dynamic peed caling cheme for an M/GI/1 PS queue we need only conider an M/M/1 FCFS queue. The ervice rate control problem in the M/M/1 FCFS queue ha been tudied extenively [12], [18], [19]. In particular, George and Harrion [12] provide an elegant olution to the problem of electing the tate-dependent proceing peed to minimize a weighted um of an arbitrary holding cot with a proceing peed cot. Specifically, the tatedependent proceing peed can be framed a the olution to a tochatic dynamic program, to which [12] provide an efficient numerical olution. In the remainder of thi ection, we will provide an overview of thi numerical approach. The core of thi approach will form the bai of our derivation of bound on the peed in Section IV. We will decribe the algorithm of [12] pecialized to the cae conidered in thi paper, where the holding cot in tate n i imply n. Further, we will generalize the decription to allow arbitrary arrival rate, λ. The olution tart with an etimate z of the minimal cot per unit time, including both the occupancy cot and the energy cot. A in [12], [19], [2], the minimum cot of returning from tate n to the empty ytem i given by the dynamic program v n = inf A { 1 λ µ [λ αβ n z ] µ λ µ v n 1 λ } λ µ v n1 where A i the et of available peed. We will uually aume A = R {}. With the ubtitution u n = λ(v n v n 1 ), thi can be written a [12], [2] u n1 = up A { z n λ α β u n ρ }. (6)

3 3 Two additional function are defined. Firt, ( ) α/(α 1) u φ(u) = up{ux/ρ λx α /β} = (α 1). (7) x A α Second, the minimum value of x which achieve thi upremum, normalized to be dimenionle, i ψ(u) = 1 { min x : ux } ( ) 1/(α 1) β1/α ρ λxα u β = φ(u) =. α (8) Given the etimate of z, u n atify u 1 = z u n1 = φ(u n ) n z. (9a) (9b) The value of z can be found a the minimum value uch that (u n ) n=1 i an increaing equence. Thi allow z to be found by an efficient binary earch, after which u n can in principle be found recurively. The peed in tate n i then given by n β 1/α = ψ(u n). (1) Thi highlight the fact that = ρ/β 1/α provide the appropriate caling of the workload information becaue the cot z, normalized peed β 1/α and variable u n depend on λ, µ and β only through. IV. BOUNDS ON OPTIMAL DYNAMIC SPEED SCALING In the prior ection, we preented the deign for gated-tatic and dynamic peed caling. In the firt cae, the peed wa preented more-or-le explicitly, however in the third cae we preented only a recurive numerical algorithm for determining the dynamic peed caling. In thi ection, we provide reult exhibiting the tructure of the dynamic peed and the performance they achieve. The main reult of thi ection are ummarized in Table I. The bound on z for arbitrary α are eentially tight (i.e., agree to leading order) in the limit of mall or large. Due to the complicated form of the general reult, we illutrate the bound for the pecific cae of α = 2 to provide inight. In particular, it i eay to ee the behavior of n and z a a function of and n in the cae of α = 2. Thi lead to intereting obervation. For example, it illutrate a connection between the tochatic policy and policie analyzed in the wort-cae model. In particular, Banal, Pruh and Stein [11] howed that, when nothing i known about future arrival, a policy that give peed of the form n = (n/(α 1)) 1/α i contant-competitive, i.e., in the wort cae the total cot i within a contant of. Thi matche the aymptotic behavior of the bound for α = 2 for large n. Thi behavior can alo be oberved for general α (Lemma 7 and Theorem 4). A. Bound on cot We tart the analyi by providing bound on z in thi ubection, and then uing the bound on z to bound n above and below (Section IV-B and IV-C). Recall that z g i the total cot under gated-tatic. Theorem 2. max ( α, α(α 1) (1/α) 1) z z g = G(; α)α 1 G(; α) Proof: The cot z i bounded above by the cot of the gated-tatic policy, which i imply z g = G(; α) G(; α)α 1. (15) Two lower bound can be obtained a follow. In order to maintain tability, the time-average peed mut atify E[] ρ. But z > E[ α ]/β (E[]) α /β by Jenen inequality and the convexity of ( ) α. Thu z > E[α ] β ρα β = α. (16) For mall load, thi bound i quite looe. Another bound come from conidering the minimum cot of proceing a ingle job of ize X, with no waiting time or proceor haring. It i to erve the job at a contant rate [2]. Thu z λ E X [ min ( X α β X )]. The right hand ide i minimized for = (β/(α 1)) 1/α independent of X, giving z ρβ 1/α α(α 1) (1/α) 1. Thu ( z max α, α(α 1) (1/α) 1). (17) The form of the bound on z are complicated, o it i ueful to look at the particular cae of α = 2. Corollary 3. For α = 2, gated-tatic ha cot within a factor of 2 of. Specifically, max( 2, 2) z z g = 2 2. (18) It i perhap urpriing that uch an idealized verion of dynamic peed caling provide uch a mall magnitude of improvement over a implitic policy uch a gated-tatic. In fact, the bound of 2 i very looe when i large or mall. Further, empirically, the maximum ratio for typical α are below 1.1 (ee Figure 2). Thu there i little to be gained by dynamic caling in term of mean cot. However, Section VI how that dynamic caling dramatically improve robutne. A econd intereting obervation about Corollary 3 i that the expected repone time under thee power aware cheme remain bounded a the arrival rate λ grow. Specifically, by (16), E[T ] = z λ E[2 /β] λ 2 µ β. Thi i a marked contrat to the tandard M/GI/1 queue.

4 4 For any α, TABLE I BOUNDS ON TOTAL COSTS AND SPEED AS A FUNCTION OF THE NUMBER n 1 OF JOBS IN THE SYSTEM. ( max α, α(α 1) (1/α) 1) z σ n G(; α) G(; α)α 1 Theorem 2 (11) ( ( 1 n σ α α min α )) 1/(α 1) σ> (σ ) (σ ) 2 Theorem 8 and 4 (12) n β 1/α where σ n atifie σn α 1 ((α 1)σ n α) n (/(G(; α) ) G(; α) α 1 For α = 2, max ( 2, 2 ) z 2 2 Corollary 3 (13) n 2 n ( n min β 2n, 3 ( ) ) 1/3 Corollarie 9 and 5 (14) 2 4 For α = 2 and n < 2, a lower bound on n reult from linear interpolation between max(/2, 1) at n = 1 and at n = 2. B. Upper bound on the dynamic peed We now move to providing upper bound on the dynamic peed caling cheme. Theorem 4. For all n and α, u n n σα α σ 2 (σ ) 2 (19) for all σ >, whence ( ( n 1 n σ α β 1/α α min α )) 1/(α 1) σ> (σ ) (σ ) 2. (2) In particular, for σ = n 1/α, which i concave in n. u n n (α 1)/α (1 (1 ) α ) 2 (21) Proof: A explained in [2], (6) can be rewritten a [ ] α u n = ρ min n /β n u n1 z. (22) n Unrolling the dynamic program (22) give a joint minimization over all n [ 1 u n = ρ min α n n/β n z n ] 1 [ ] ρ min α n1 n1 /β (n 1) z u n2 n1 i = min ρ ( α i /β i z). (23) i,i n i=n j=n j An upper bound can be found by taking any (poibly ub) choice of ni for i 1, and bounding the z. Taking i = σβ 1/α > for all i n give ( ) j u n min (σ α (n j) z) σ> σ σ j= [ n σ α ] z = min σ> σ (σ ) 2. n Since z α from (17), equation (19) follow. With (1), thi etablihe (2). For n =, (21) hold ince u =. Otherwie, it follow from the inequality σ α = n(1 n 1/α ) α n(1 ) α and the fact that n 2/α 1. By pecializing to the cae when α = 2, we can provide ome intuition for the upper bound on the peed. Factoring the difference of quare in the firt term of (19) yeild one increaing term and two decreaing term. Minimizing pair of thee term give the following upper bound on u n. Corollary 5. For α = 2, n β ( n min 1/α 2n, 3 ( ) ) 1/3. (24) 2 4 C. Lower bound on the dynamic peed Finally, we prove lower bound on the dynamic peed caling cheme. We begin by bounding the peed ued when there i one job in the ytem. The following reult i an immediate conequence of Corollary 3 and (9a). Corollary 6. For α = 2, ( ) max 2, (25) β 2 Oberve that the bound in (25), like thoe in Corollary 3, are eentially tight for both large and mall, but looe for near 1, epecially the lower bound. In conjunction with (21) and (1), the following lemma how that peed choen to perform well in the wort-cae are aymptotically (for large n) in the tochatic model. Lemma 7. For ufficiently large n, ( ) 1/α n n β >. (26) 1/α α 1 The following tighter bound on the peed i obtained by uing u n u n1 and (15). Theorem 8. The caled peed σ n = n/β 1/α atifie ( (α 1)σn α ) n σ α 1 n G(; α) G(; α)α 1.

5 5 rate/qrt(β) tatic leep 1 tatic bound occupancy, n (a) = 1 rate/qrt(β) tatic leep tatic bound occupancy, n (b) = 1 Fig. 1. Rate v n, for α = 2 and different energy-aware-load,. For α = 2, thi become: Corollary 9. For α = 2 and any n 2, n β 1/α n 2. (27) Thi prove that the mode n = min n { n ρ} atifie n 2. By the following lemma, linear interpolation between max(/2, 1) and give a lower bound on n for n < 2. Lemma 1. The equence u n i trictly concave increaing. V. COMPARING STATIC AND DYNAMIC SCHEMES To thi point, we have only provided analytic reult. We now ue numerical experiment to contrat tatic and dynamic cheme. In addition, thee experiment will illutrate the tightne of the bound proven in Section IV on the dynamic peed caling cheme. We will tart by contrating the peed under each of the cheme. Figure 1 compare the dynamic peed with the tatic peed. Note that the bound on the dynamic peed are quite tight, epecially when the number of job in the ytem, n, i large. For reference, the mode of the occupancy ditribution are about 1 and 5, cloe to the point at which the peed matche the tatic peed. Note alo that the rate grow only lowly for n much larger than the typical occupancy. Thi i important ince the range over which DVS i poible i limited [15]. Although the peed of the cheme differ ignificantly from that of gated-tatic, the actual cot are very imilar, a predicted by the remark after Corollary 3. Thi i hown in Figure 2. The bound on the peed are alo very tight, both for large and mall. Part (a) how that the lower bound i looet for intermediate, where the weight given to power and repone time are comparable. Part (b) how that gated-tatic (i.e., the upper bound) ha very cloe to the cot. In addition to comparing the total cot of the cheme, it i important to contrat the mean repone time and mean energy uage. Figure 3 how the breakdown. A reference load of ρ = 3 with delay-averion β = 1 and power caling α = 2 wa compared againt changing ρ for fixed, changing β for fixed ρ and changing α. Note = 3 wa choen to maximize the ratio of z g /z. The econd cenario how that when i held fixed, but the load ρ i reduced and delay-averion cot per job, z/λ tatic leep tatic bound (a) Abolute cot, α = 2 z / z α=1.6 α=2 α= (b) Ratio of cot for gated-tatic to, z g/z. Fig. 2. Cot z v energy-aware-load. i reduced commenurately, the energy conumption become negligible. VI. ROBUST POWER-AWARE DESIGN We have een both analytically and numerically that (idealized) dynamic peed caling only marginally reduce the cot compared to the imple gated-tatic. Thi raie the quetion of whether dynamic caling i worth the complexity. Thi ection illutrate one reaon: robutne. Specifically, dynamic cheme provide ignificantly better performance in the face of burty traffic and mi-etimation of workload. We focu on robutne with repect to the load, ρ. The peed are enitive to ρ, but in reality thi parameter mut be etimated, and will be time-varying. It i eay to ee the problem mi-etimation of ρ caue for tatic peed deign. If the load i not known, then the elected peed mut be atifactory for all poible anticipated load. Conider the cae that it i only known that ρ [ρ, ρ]. Let z(ρ 1 ρ 2 ) denote the expected cot per unit time if the arrival rate i ρ 1, but the peed wa optimized for ρ 2. Then, the robut deign problem i to elect the peed ρ uch that min ρ max ρ [ρ, ρ] z(ρ ρ ). The deign i to proviion for the highet foreeen load, i.e., max ρ [ρ, ρ] z(ρ ρ ) = z( ρ ρ ). However, thi i wateful in the typical cae that the load i le than ρ. The fragility of tatic peed deign i illutrated in Figure 4, which how that when peed i underproviioned, the erver i untable, and when it i overproviioned the deign i wateful. Optimal dynamic caling i not immune to mi-etimation of ρ, ince n i highly dependent on ρ. However, becaue the peed adapt to the queue length, dynamic caling i more robut. Figure 4 how thi improvement. Though the dynamic cheme i more robut than a tatic cheme, robutne can be improved further. Specifically, conider the following peed caling cheme that we term linear. It cale the erver peed in proportion to the queue length, i.e., n /β 1/α = n. Figure 4 how that the linear caling provide ignificantly improved robutne when compared with the dynamic cheme; indeed, the cheme i only for deign with ρ [7, 14]. Further, when ρ i in thi region, the linear caling provide only lightly higher cot than the caling. The price that linear caling pay i that it require very high proceing peed

6 6 Fig. 3. delay or energy (normalized unit) energy repone time = 3 = 3 α = 2 α = 2 ρ = 3 ρ =.3 β = 1 β=.1 Optimal Static leep Static = 3 = 3 α = 2 α = 2 ρ = 3 ρ = 3 β=.1 β = 1 Breakdown of E[T ] and E[ α ], for everal cenario. when the occupancy i high, which may not be upported by the hardware. In addition to the numerical illutration above, we can compare robutne analytically in the cae of α = 2. Theorem 11 how that the cot of the linear cheme i exactly the ame a the cot of the gated-tatic cheme when ρ i known exactly. Thu, the cot of the linear cheme i within a factor of 2 of, even without uing information about ρ. Theorem 11. When α = 2, z g = z lin. Thu, z lin 2z. Theorem 12. Conider a ytem deigned for target load ρ that i operating at load ρ. When α = 2, z lin = ρ2 β 2 ρ (28) β z = z lin ρ ( ) ɛ 2. (29) β β ɛ VII. CONCLUDING REMARKS Speed caling i an important method for reducing energy conumption in computer communication ytem. Intrinically, it trade off the mean repone time and the mean energy conumption, and thi paper provide inight into thi tradeoff uing a tochatic analyi. Specifically, in the M/GI/1 PS model, both bound and aymptotic for the peed caling cheme are provided. Thee bound are tight for mall and large and provide a number of inight, e.g., that the mean repone time i bounded a the load grow under the dynamic peed caling and that the dynamic peed in the tochatic model match (for large n) dynamic peed caling that have been hown to have good wort-cae performance. Surpriingly, the bound alo illutrate that a imple cheme which run at peed when the ytem i idle and ue a tatic rate while the ytem i buy provide performance within a factor of 2 of the dynamic peed caling. However, the value of dynamic peed caling i alo illutrated dynamic peed caling cheme provide ignificantly improved robutne to burty traffic and mi-etimation of workload parameter. Interetingly, the dynamic cheme that optimize the mean cot i no longer when robutne i conidered. In particular, a cheme that cale peed linearly with n can provide ignificantly improved robutne while increaing cot only lightly. cot, z tatic leep linear peed deign ρ Fig. 4. Cot at load ρ = 1, when peed are deigned for deign ρ, uing β = 1, α = 2. REFERENCES [1] J. Baliga, R. Ayre, W. Sorin, K. Hinton, and R. Tucker, Energy conumption in acce network, in IEEE Conf. Optical Fiber Communication (OFC), Feb. 28, pp [2] F. Yao, A. Demer, and S. Shenker, A cheduling model for reduced CPU energy, in Proc. IEEE Symp. Foundation of Computer Science (FOCS), 1995, pp [3] N. Banal, T. Kimbrel, and K. Pruh, Speed caling to manage energy and temperature, J. ACM, vol. 54, no. 1, pp. 1 39, Mar. 27. [4] Y. Zhu and F. Mueller, Feedback EDF cheduling of real-time tak exploiting dynamic voltage caling, Real Time Sytem, vol. 31, pp , Dec. 25. [5] L. Yuan and G. Qu, Analyi of energy reduction on dynamic voltage caling-enabled ytem, IEEE Tran. Comput.-Aided De. Integr. Circuit Syt., vol. 24, no. 12, pp , Dec. 25. [6] S. Herbert and D. Marculecu, Analyi of dynamic voltage/frequency caling in chip-multiproceor, in Proc. ISLPED, 27, p. 6. [7] S. Zhang and K. S. Catha, Approximation algorithm for the temperature-aware cheduling problem, in Proc. IEEE Int. Conf. Comp. Aided Deign, Nov. 27, pp [8] K. Pruh, R. van Stee, and P. Uthaiombut, Speed caling of tak with precedence contraint, in Proc. Workhop on Approximation and Online Algorithm, 25. [9] D. P. Bunde, Power-aware cheduling for makepand and flow, in Proc. ACM Symp. Parallel Alg. and Arch., 26. [1] S. Alber and H. Fujiwara, Energy-efficient algorithm for flow time minimization, in Lecture Note in Computer Science (STACS), vol. 3884, 26, pp [11] N. Banal, K. Pruh, and C. Stein, Speed caling for weighted flow time, in Proc. ACM-SIAM SODA, 27, pp [12] J. M. George and J. M. Harrion, Dynamic control of a queue with adjutable ervice rate, Operation Reearch, vol. 49, no. 5, pp , Sep. 21. [13] J. O. Kephart, H. Chan, R. Da, D. W. Levine, G. Teauro, F. Rawon, and C. Lefurgy, Coordinating multiple autonomic manager to achieve pecified power-performance tradeoff, in Proc. Int Conf Autonomic Computing, 27, p. 24. [14] A. Wierman, L. L. H. Andrew, and A. Tang, Power-aware peed caling in proceor haring ytem. [Online]. Available: adamw/paper/power-2.pdf [15] S. Kaxira and M. Martonoi, Computer Architecture Technique for Power-Efficiency. Morgan and Claypool, 28. [16] O. S. Unal and I. Koren, Sytem-level power-aware deign technique in real-time ytem, Proc. IEEE, vol. 91, no. 7, pp , 23. [17] F. P. Kelly, Reveribility and Stochatic Network. Wiley, [18] B. Ata and S. Shneoron, Dynamic control of an M/M/1 ervice ytem with adjutable arrival and ervice rate, Management Science, vol. 51, no. 11, pp , Nov. 26. [19] D. Low, Optimal pricing policie for an M/M/ queue, Operation Reearch, vol. 22, pp , [2] J. Wijngaard and S. Stidham, Forward recurion of Markov deciion procee with kip-free-to-the-right tranition, part I: Theory and algorithm, Mathematic of Operation Reearch, vol. 11, no. 2, pp , May VIII. ACKNOWLEDGEMENT Thi work wa upported by grant from NSF CCF and CNS 43552, Microoft Reearch and the Lee Center for Advanced Networking.

Proactive Serving Decreases User Delay Exponentially: The Light-tailed Service Time Case

Proactive Serving Decreases User Delay Exponentially: The Light-tailed Service Time Case Proactive Serving Decreae Uer Delay Exponentially: The Light-tailed Service Time Cae Shaoquan Zhang, Longbo Huang, Minghua Chen, and Xin Liu Abtract In online ervice ytem, the delay experienced by uer

More information

Preemptive scheduling on a small number of hierarchical machines

Preemptive scheduling on a small number of hierarchical machines Available online at www.ciencedirect.com Information and Computation 06 (008) 60 619 www.elevier.com/locate/ic Preemptive cheduling on a mall number of hierarchical machine György Dóa a, Leah Eptein b,

More information

Gain and Phase Margins Based Delay Dependent Stability Analysis of Two- Area LFC System with Communication Delays

Gain and Phase Margins Based Delay Dependent Stability Analysis of Two- Area LFC System with Communication Delays Gain and Phae Margin Baed Delay Dependent Stability Analyi of Two- Area LFC Sytem with Communication Delay Şahin Sönmez and Saffet Ayaun Department of Electrical Engineering, Niğde Ömer Halidemir Univerity,

More information

ON THE APPROXIMATION ERROR IN HIGH DIMENSIONAL MODEL REPRESENTATION. Xiaoqun Wang

ON THE APPROXIMATION ERROR IN HIGH DIMENSIONAL MODEL REPRESENTATION. Xiaoqun Wang Proceeding of the 2008 Winter Simulation Conference S. J. Maon, R. R. Hill, L. Mönch, O. Roe, T. Jefferon, J. W. Fowler ed. ON THE APPROXIMATION ERROR IN HIGH DIMENSIONAL MODEL REPRESENTATION Xiaoqun Wang

More information

Codes Correcting Two Deletions

Codes Correcting Two Deletions 1 Code Correcting Two Deletion Ryan Gabry and Frederic Sala Spawar Sytem Center Univerity of California, Lo Angele ryan.gabry@navy.mil fredala@ucla.edu Abtract In thi work, we invetigate the problem of

More information

Alternate Dispersion Measures in Replicated Factorial Experiments

Alternate Dispersion Measures in Replicated Factorial Experiments Alternate Diperion Meaure in Replicated Factorial Experiment Neal A. Mackertich The Raytheon Company, Sudbury MA 02421 Jame C. Benneyan Northeatern Univerity, Boton MA 02115 Peter D. Krau The Raytheon

More information

Stochastic Optimization with Inequality Constraints Using Simultaneous Perturbations and Penalty Functions

Stochastic Optimization with Inequality Constraints Using Simultaneous Perturbations and Penalty Functions Stochatic Optimization with Inequality Contraint Uing Simultaneou Perturbation and Penalty Function I-Jeng Wang* and Jame C. Spall** The John Hopkin Univerity Applied Phyic Laboratory 11100 John Hopkin

More information

Optimal revenue management in two class pre-emptive delay dependent Markovian queues

Optimal revenue management in two class pre-emptive delay dependent Markovian queues Optimal revenue management in two cla pre-emptive delay dependent Markovian queue Manu K. Gupta, N. Hemachandra and J. Venkatewaran Indutrial Engineering and Operation Reearch, IIT Bombay March 15, 2015

More information

One Class of Splitting Iterative Schemes

One Class of Splitting Iterative Schemes One Cla of Splitting Iterative Scheme v Ciegi and V. Pakalnytė Vilniu Gedimina Technical Univerity Saulėtekio al. 11, 2054, Vilniu, Lithuania rc@fm.vtu.lt Abtract. Thi paper deal with the tability analyi

More information

Clustering Methods without Given Number of Clusters

Clustering Methods without Given Number of Clusters Clutering Method without Given Number of Cluter Peng Xu, Fei Liu Introduction A we now, mean method i a very effective algorithm of clutering. It mot powerful feature i the calability and implicity. However,

More information

CONGESTION control is a key functionality in modern

CONGESTION control is a key functionality in modern IEEE TRANSACTIONS ON INFORMATION TEORY, VOL. X, NO. X, XXXXXXX 2008 On the Connection-Level Stability of Congetion-Controlled Communication Network Xiaojun Lin, Member, IEEE, Ne B. Shroff, Fellow, IEEE,

More information

Evolutionary Algorithms Based Fixed Order Robust Controller Design and Robustness Performance Analysis

Evolutionary Algorithms Based Fixed Order Robust Controller Design and Robustness Performance Analysis Proceeding of 01 4th International Conference on Machine Learning and Computing IPCSIT vol. 5 (01) (01) IACSIT Pre, Singapore Evolutionary Algorithm Baed Fixed Order Robut Controller Deign and Robutne

More information

Performance Evaluation

Performance Evaluation Performance Evaluation 95 (206) 40 Content lit available at ScienceDirect Performance Evaluation journal homepage: www.elevier.com/locate/peva Optimal cheduling in call center with a callback option Benjamin

More information

Theoretical Computer Science. Optimal algorithms for online scheduling with bounded rearrangement at the end

Theoretical Computer Science. Optimal algorithms for online scheduling with bounded rearrangement at the end Theoretical Computer Science 4 (0) 669 678 Content lit available at SciVere ScienceDirect Theoretical Computer Science journal homepage: www.elevier.com/locate/tc Optimal algorithm for online cheduling

More information

Suggested Answers To Exercises. estimates variability in a sampling distribution of random means. About 68% of means fall

Suggested Answers To Exercises. estimates variability in a sampling distribution of random means. About 68% of means fall Beyond Significance Teting ( nd Edition), Rex B. Kline Suggeted Anwer To Exercie Chapter. The tatitic meaure variability among core at the cae level. In a normal ditribution, about 68% of the core fall

More information

The Impact of Imperfect Scheduling on Cross-Layer Rate. Control in Multihop Wireless Networks

The Impact of Imperfect Scheduling on Cross-Layer Rate. Control in Multihop Wireless Networks The mpact of mperfect Scheduling on Cro-Layer Rate Control in Multihop Wirele Network Xiaojun Lin and Ne B. Shroff Center for Wirele Sytem and Application (CWSA) School of Electrical and Computer Engineering,

More information

White Rose Research Online URL for this paper: Version: Accepted Version

White Rose Research Online URL for this paper:   Version: Accepted Version Thi i a repoitory copy of Identification of nonlinear ytem with non-peritent excitation uing an iterative forward orthogonal leat quare regreion algorithm. White Roe Reearch Online URL for thi paper: http://eprint.whiteroe.ac.uk/107314/

More information

Advanced D-Partitioning Analysis and its Comparison with the Kharitonov s Theorem Assessment

Advanced D-Partitioning Analysis and its Comparison with the Kharitonov s Theorem Assessment Journal of Multidiciplinary Engineering Science and Technology (JMEST) ISSN: 59- Vol. Iue, January - 5 Advanced D-Partitioning Analyi and it Comparion with the haritonov Theorem Aement amen M. Yanev Profeor,

More information

OVERFLOW PROBABILITY IN AN ATM QUEUE WITH SELF-SIMILAR INPUT TRAFFIC

OVERFLOW PROBABILITY IN AN ATM QUEUE WITH SELF-SIMILAR INPUT TRAFFIC Copyright by IEEE OVERFLOW PROBABILITY IN AN ATM QUEUE WITH SELF-SIMILAR INPUT TRAFFIC Bori Tybakov Intitute for Problem in Information Tranmiion Ruian Academy of Science Mocow, Ruia e-mail: bt@ippi ac

More information

Control Systems Analysis and Design by the Root-Locus Method

Control Systems Analysis and Design by the Root-Locus Method 6 Control Sytem Analyi and Deign by the Root-Locu Method 6 1 INTRODUCTION The baic characteritic of the tranient repone of a cloed-loop ytem i cloely related to the location of the cloed-loop pole. If

More information

Social Studies 201 Notes for November 14, 2003

Social Studies 201 Notes for November 14, 2003 1 Social Studie 201 Note for November 14, 2003 Etimation of a mean, mall ample ize Section 8.4, p. 501. When a reearcher ha only a mall ample ize available, the central limit theorem doe not apply to the

More information

EXTENDED STABILITY MARGINS ON CONTROLLER DESIGN FOR NONLINEAR INPUT DELAY SYSTEMS. Otto J. Roesch, Hubert Roth, Asif Iqbal

EXTENDED STABILITY MARGINS ON CONTROLLER DESIGN FOR NONLINEAR INPUT DELAY SYSTEMS. Otto J. Roesch, Hubert Roth, Asif Iqbal EXTENDED STABILITY MARGINS ON CONTROLLER DESIGN FOR NONLINEAR INPUT DELAY SYSTEMS Otto J. Roech, Hubert Roth, Aif Iqbal Intitute of Automatic Control Engineering Univerity Siegen, Germany {otto.roech,

More information

into a discrete time function. Recall that the table of Laplace/z-transforms is constructed by (i) selecting to get

into a discrete time function. Recall that the table of Laplace/z-transforms is constructed by (i) selecting to get Lecture 25 Introduction to Some Matlab c2d Code in Relation to Sampled Sytem here are many way to convert a continuou time function, { h( t) ; t [0, )} into a dicrete time function { h ( k) ; k {0,,, }}

More information

Simple Observer Based Synchronization of Lorenz System with Parametric Uncertainty

Simple Observer Based Synchronization of Lorenz System with Parametric Uncertainty IOSR Journal of Electrical and Electronic Engineering (IOSR-JEEE) ISSN: 78-676Volume, Iue 6 (Nov. - Dec. 0), PP 4-0 Simple Oberver Baed Synchronization of Lorenz Sytem with Parametric Uncertainty Manih

More information

To provide useful practical insight into the performance of service-oriented (non-revenue-generating) call

To provide useful practical insight into the performance of service-oriented (non-revenue-generating) call MANAGEMENT SCIENCE Vol. 50, No. 10, October 2004, pp. 1449 1461 in 0025-1909 ein 1526-5501 04 5010 1449 inform doi 10.1287/mnc.1040.0279 2004 INFORMS Efficiency-Driven Heavy-Traffic Approximation for Many-Server

More information

Source slideplayer.com/fundamentals of Analytical Chemistry, F.J. Holler, S.R.Crouch. Chapter 6: Random Errors in Chemical Analysis

Source slideplayer.com/fundamentals of Analytical Chemistry, F.J. Holler, S.R.Crouch. Chapter 6: Random Errors in Chemical Analysis Source lideplayer.com/fundamental of Analytical Chemitry, F.J. Holler, S.R.Crouch Chapter 6: Random Error in Chemical Analyi Random error are preent in every meaurement no matter how careful the experimenter.

More information

CHEAP CONTROL PERFORMANCE LIMITATIONS OF INPUT CONSTRAINED LINEAR SYSTEMS

CHEAP CONTROL PERFORMANCE LIMITATIONS OF INPUT CONSTRAINED LINEAR SYSTEMS Copyright 22 IFAC 5th Triennial World Congre, Barcelona, Spain CHEAP CONTROL PERFORMANCE LIMITATIONS OF INPUT CONSTRAINED LINEAR SYSTEMS Tritan Pérez Graham C. Goodwin Maria M. Serón Department of Electrical

More information

Factor Analysis with Poisson Output

Factor Analysis with Poisson Output Factor Analyi with Poion Output Gopal Santhanam Byron Yu Krihna V. Shenoy, Department of Electrical Engineering, Neurocience Program Stanford Univerity Stanford, CA 94305, USA {gopal,byronyu,henoy}@tanford.edu

More information

Optimal Coordination of Samples in Business Surveys

Optimal Coordination of Samples in Business Surveys Paper preented at the ICES-III, June 8-, 007, Montreal, Quebec, Canada Optimal Coordination of Sample in Buine Survey enka Mach, Ioana Şchiopu-Kratina, Philip T Rei, Jean-Marc Fillion Statitic Canada New

More information

On the Stability Region of Congestion Control

On the Stability Region of Congestion Control On the Stability Region of Congetion Control Xiaojun Lin and Ne B. Shroff School of Electrical and Computer Engineering Purdue Univerity, Wet Lafayette, IN 47906 {linx,hroff}@ecn.purdue.edu Abtract It

More information

To describe a queuing system, an input process and an output process has to be specified.

To describe a queuing system, an input process and an output process has to be specified. 5. Queue (aiting Line) Queuing terminology Input Service Output To decribe a ueuing ytem, an input proce and an output proce ha to be pecified. For example ituation input proce output proce Bank Cutomer

More information

State Space: Observer Design Lecture 11

State Space: Observer Design Lecture 11 State Space: Oberver Deign Lecture Advanced Control Sytem Dr Eyad Radwan Dr Eyad Radwan/ACS/ State Space-L Controller deign relie upon acce to the tate variable for feedback through adjutable gain. Thi

More information

Lecture 10 Filtering: Applied Concepts

Lecture 10 Filtering: Applied Concepts Lecture Filtering: Applied Concept In the previou two lecture, you have learned about finite-impule-repone (FIR) and infinite-impule-repone (IIR) filter. In thee lecture, we introduced the concept of filtering

More information

Digital Control System

Digital Control System Digital Control Sytem - A D D A Micro ADC DAC Proceor Correction Element Proce Clock Meaurement A: Analog D: Digital Continuou Controller and Digital Control Rt - c Plant yt Continuou Controller Digital

More information

USING NONLINEAR CONTROL ALGORITHMS TO IMPROVE THE QUALITY OF SHAKING TABLE TESTS

USING NONLINEAR CONTROL ALGORITHMS TO IMPROVE THE QUALITY OF SHAKING TABLE TESTS October 12-17, 28, Beijing, China USING NONLINEAR CONTR ALGORITHMS TO IMPROVE THE QUALITY OF SHAKING TABLE TESTS T.Y. Yang 1 and A. Schellenberg 2 1 Pot Doctoral Scholar, Dept. of Civil and Env. Eng.,

More information

Lecture 4 Topic 3: General linear models (GLMs), the fundamentals of the analysis of variance (ANOVA), and completely randomized designs (CRDs)

Lecture 4 Topic 3: General linear models (GLMs), the fundamentals of the analysis of variance (ANOVA), and completely randomized designs (CRDs) Lecture 4 Topic 3: General linear model (GLM), the fundamental of the analyi of variance (ANOVA), and completely randomized deign (CRD) The general linear model One population: An obervation i explained

More information

IEOR 3106: Fall 2013, Professor Whitt Topics for Discussion: Tuesday, November 19 Alternating Renewal Processes and The Renewal Equation

IEOR 3106: Fall 2013, Professor Whitt Topics for Discussion: Tuesday, November 19 Alternating Renewal Processes and The Renewal Equation IEOR 316: Fall 213, Profeor Whitt Topic for Dicuion: Tueday, November 19 Alternating Renewal Procee and The Renewal Equation 1 Alternating Renewal Procee An alternating renewal proce alternate between

More information

A Simplified Methodology for the Synthesis of Adaptive Flight Control Systems

A Simplified Methodology for the Synthesis of Adaptive Flight Control Systems A Simplified Methodology for the Synthei of Adaptive Flight Control Sytem J.ROUSHANIAN, F.NADJAFI Department of Mechanical Engineering KNT Univerity of Technology 3Mirdamad St. Tehran IRAN Abtract- A implified

More information

CONTROL SYSTEMS, ROBOTICS AND AUTOMATION Vol. VIII Decoupling Control - M. Fikar

CONTROL SYSTEMS, ROBOTICS AND AUTOMATION Vol. VIII Decoupling Control - M. Fikar DECOUPLING CONTROL M. Fikar Department of Proce Control, Faculty of Chemical and Food Technology, Slovak Univerity of Technology in Bratilava, Radlinkého 9, SK-812 37 Bratilava, Slovakia Keyword: Decoupling:

More information

Reliability Analysis of Embedded System with Different Modes of Failure Emphasizing Reboot Delay

Reliability Analysis of Embedded System with Different Modes of Failure Emphasizing Reboot Delay International Journal of Applied Science and Engineering 3., 4: 449-47 Reliability Analyi of Embedded Sytem with Different Mode of Failure Emphaizing Reboot Delay Deepak Kumar* and S. B. Singh Department

More information

A Simple Approach to Synthesizing Naïve Quantized Control for Reference Tracking

A Simple Approach to Synthesizing Naïve Quantized Control for Reference Tracking A Simple Approach to Syntheizing Naïve Quantized Control for Reference Tracking SHIANG-HUA YU Department of Electrical Engineering National Sun Yat-Sen Univerity 70 Lien-Hai Road, Kaohiung 804 TAIAN Abtract:

More information

LOW ORDER MIMO CONTROLLER DESIGN FOR AN ENGINE DISTURBANCE REJECTION PROBLEM. P.Dickinson, A.T.Shenton

LOW ORDER MIMO CONTROLLER DESIGN FOR AN ENGINE DISTURBANCE REJECTION PROBLEM. P.Dickinson, A.T.Shenton LOW ORDER MIMO CONTROLLER DESIGN FOR AN ENGINE DISTURBANCE REJECTION PROBLEM P.Dickinon, A.T.Shenton Department of Engineering, The Univerity of Liverpool, Liverpool L69 3GH, UK Abtract: Thi paper compare

More information

Stochastic Perishable Inventory Control in a Service Facility System Maintaining Inventory for Service: Semi Markov Decision Problem

Stochastic Perishable Inventory Control in a Service Facility System Maintaining Inventory for Service: Semi Markov Decision Problem Stochatic Perihable Inventory Control in a Service Facility Sytem Maintaining Inventory for Service: Semi Markov Deciion Problem R.Mugeh 1,S.Krihnakumar 2, and C.Elango 3 1 mugehrengawamy@gmail.com 2 krihmathew@gmail.com

More information

Competitive Analysis of Task Scheduling Algorithms on a Fault-Prone Machine and the Impact of Resource Augmentation

Competitive Analysis of Task Scheduling Algorithms on a Fault-Prone Machine and the Impact of Resource Augmentation Competitive Analyi of Tak Scheduling Algorithm on a Fault-Prone Machine and the Impact of Reource Augmentation Antonio Fernández Anta a, Chryi Georgiou b, Dariuz R. Kowalki c, Elli Zavou a,d,1 a Intitute

More information

Random vs. Deterministic Deployment of Sensors in the Presence of Failures and Placement Errors

Random vs. Deterministic Deployment of Sensors in the Presence of Failures and Placement Errors Random v. Determinitic Deployment of Senor in the Preence of Failure and Placement Error Paul Baliter Univerity of Memphi pbalitr@memphi.edu Santoh Kumar Univerity of Memphi antoh.kumar@memphi.edu Abtract

More information

A BATCH-ARRIVAL QUEUE WITH MULTIPLE SERVERS AND FUZZY PARAMETERS: PARAMETRIC PROGRAMMING APPROACH

A BATCH-ARRIVAL QUEUE WITH MULTIPLE SERVERS AND FUZZY PARAMETERS: PARAMETRIC PROGRAMMING APPROACH Mathematical and Computational Application Vol. 11 No. pp. 181-191 006. Aociation for Scientific Reearch A BATCH-ARRIVA QEE WITH MTIPE SERVERS AND FZZY PARAMETERS: PARAMETRIC PROGRAMMING APPROACH Jau-Chuan

More information

Lecture 21. The Lovasz splitting-off lemma Topics in Combinatorial Optimization April 29th, 2004

Lecture 21. The Lovasz splitting-off lemma Topics in Combinatorial Optimization April 29th, 2004 18.997 Topic in Combinatorial Optimization April 29th, 2004 Lecture 21 Lecturer: Michel X. Goeman Scribe: Mohammad Mahdian 1 The Lovaz plitting-off lemma Lovaz plitting-off lemma tate the following. Theorem

More information

Real-Time Identification of Sliding Friction Using LabVIEW FPGA

Real-Time Identification of Sliding Friction Using LabVIEW FPGA Real-Time Identification of Sliding Friction Uing LabVIEW FPGA M. Laine Mear, Jeannie S. Falcon, IEEE, and Thoma R. Kurfe, IEEE Abtract Friction i preent in all mechanical ytem, and can greatly affect

More information

Problem Set 8 Solutions

Problem Set 8 Solutions Deign and Analyi of Algorithm April 29, 2015 Maachuett Intitute of Technology 6.046J/18.410J Prof. Erik Demaine, Srini Devada, and Nancy Lynch Problem Set 8 Solution Problem Set 8 Solution Thi problem

More information

Call Centers with a Postponed Callback Offer

Call Centers with a Postponed Callback Offer Call Center with a Potponed Callback Offer Benjamin Legro, Sihan Ding, Rob Van Der Mei, Oualid Jouini To cite thi verion: Benjamin Legro, Sihan Ding, Rob Van Der Mei, Oualid Jouini. Call Center with a

More information

CHAPTER 4 DESIGN OF STATE FEEDBACK CONTROLLERS AND STATE OBSERVERS USING REDUCED ORDER MODEL

CHAPTER 4 DESIGN OF STATE FEEDBACK CONTROLLERS AND STATE OBSERVERS USING REDUCED ORDER MODEL 98 CHAPTER DESIGN OF STATE FEEDBACK CONTROLLERS AND STATE OBSERVERS USING REDUCED ORDER MODEL INTRODUCTION The deign of ytem uing tate pace model for the deign i called a modern control deign and it i

More information

A Constraint Propagation Algorithm for Determining the Stability Margin. The paper addresses the stability margin assessment for linear systems

A Constraint Propagation Algorithm for Determining the Stability Margin. The paper addresses the stability margin assessment for linear systems A Contraint Propagation Algorithm for Determining the Stability Margin of Linear Parameter Circuit and Sytem Lubomir Kolev and Simona Filipova-Petrakieva Abtract The paper addree the tability margin aement

More information

Bogoliubov Transformation in Classical Mechanics

Bogoliubov Transformation in Classical Mechanics Bogoliubov Tranformation in Claical Mechanic Canonical Tranformation Suppoe we have a et of complex canonical variable, {a j }, and would like to conider another et of variable, {b }, b b ({a j }). How

More information

Chapter Landscape of an Optimization Problem. Local Search. Coping With NP-Hardness. Gradient Descent: Vertex Cover

Chapter Landscape of an Optimization Problem. Local Search. Coping With NP-Hardness. Gradient Descent: Vertex Cover Coping With NP-Hardne Chapter 12 Local Search Q Suppoe I need to olve an NP-hard problem What hould I do? A Theory ay you're unlikely to find poly-time algorithm Mut acrifice one of three deired feature

More information

Convex Hulls of Curves Sam Burton

Convex Hulls of Curves Sam Burton Convex Hull of Curve Sam Burton 1 Introduction Thi paper will primarily be concerned with determining the face of convex hull of curve of the form C = {(t, t a, t b ) t [ 1, 1]}, a < b N in R 3. We hall

More information

An Inequality for Nonnegative Matrices and the Inverse Eigenvalue Problem

An Inequality for Nonnegative Matrices and the Inverse Eigenvalue Problem An Inequality for Nonnegative Matrice and the Invere Eigenvalue Problem Robert Ream Program in Mathematical Science The Univerity of Texa at Dalla Box 83688, Richardon, Texa 7583-688 Abtract We preent

More information

Thermal Resistance Measurements and Thermal Transient Analysis of Power Chip Slug-Up and Slug-Down Mounted on HDI Substrate

Thermal Resistance Measurements and Thermal Transient Analysis of Power Chip Slug-Up and Slug-Down Mounted on HDI Substrate Intl Journal of Microcircuit and Electronic Packaging Thermal Reitance Meaurement and Thermal Tranient Analyi of Power Chip Slug-Up and Slug-Down Mounted on HDI Subtrate Claudio Sartori Magneti Marelli

More information

Real-time identification of sliding friction using LabVIEW FPGA

Real-time identification of sliding friction using LabVIEW FPGA Clemon Univerity TigerPrint Publication Automotive Engineering 6-26 Real-time identification of liding friction uing LabVIEW FPGA Laine Mear Clemon Univerity, mear@clemon.edu Jeannie S. Falcon IEEE Thoma

More information

A Geometric Characterization of the Power of Finite Adaptability in Multi-stage Stochastic and Adaptive Optimization

A Geometric Characterization of the Power of Finite Adaptability in Multi-stage Stochastic and Adaptive Optimization MATHEMATICS OF OPERATIONS RESEARCH Vol. xx, No. x, Xxxxxxx 200x, pp. xxx xxx ISSN 0364-765X EISSN 526-547 0x xx0x 0xxx inform DOI 0.287/moor.xxxx.xxxx c 200x INFORMS A Geometric Characterization of the

More information

Chapter 13. Root Locus Introduction

Chapter 13. Root Locus Introduction Chapter 13 Root Locu 13.1 Introduction In the previou chapter we had a glimpe of controller deign iue through ome imple example. Obviouly when we have higher order ytem, uch imple deign technique will

More information

New bounds for Morse clusters

New bounds for Morse clusters New bound for More cluter Tamá Vinkó Advanced Concept Team, European Space Agency, ESTEC Keplerlaan 1, 2201 AZ Noordwijk, The Netherland Tama.Vinko@ea.int and Arnold Neumaier Fakultät für Mathematik, Univerität

More information

III.9. THE HYSTERESIS CYCLE OF FERROELECTRIC SUBSTANCES

III.9. THE HYSTERESIS CYCLE OF FERROELECTRIC SUBSTANCES III.9. THE HYSTERESIS CYCLE OF FERROELECTRIC SBSTANCES. Work purpoe The analyi of the behaviour of a ferroelectric ubtance placed in an eternal electric field; the dependence of the electrical polariation

More information

Performance Bounds for Constrained Linear Min-Max Control

Performance Bounds for Constrained Linear Min-Max Control 01 European Control Conference (ECC) July 1-19, 01, Zürich, Switzerland. Performance Bound for Contrained Linear Min-Max Control Tyler H. Summer and Paul J. Goulart Abtract Thi paper propoe a method to

More information

Chapter 4. The Laplace Transform Method

Chapter 4. The Laplace Transform Method Chapter 4. The Laplace Tranform Method The Laplace Tranform i a tranformation, meaning that it change a function into a new function. Actually, it i a linear tranformation, becaue it convert a linear combination

More information

7.2 INVERSE TRANSFORMS AND TRANSFORMS OF DERIVATIVES 281

7.2 INVERSE TRANSFORMS AND TRANSFORMS OF DERIVATIVES 281 72 INVERSE TRANSFORMS AND TRANSFORMS OF DERIVATIVES 28 and i 2 Show how Euler formula (page 33) can then be ued to deduce the reult a ( a) 2 b 2 {e at co bt} {e at in bt} b ( a) 2 b 2 5 Under what condition

More information

Calculation of the temperature of boundary layer beside wall with time-dependent heat transfer coefficient

Calculation of the temperature of boundary layer beside wall with time-dependent heat transfer coefficient Ŕ periodica polytechnica Mechanical Engineering 54/1 21 15 2 doi: 1.3311/pp.me.21-1.3 web: http:// www.pp.bme.hu/ me c Periodica Polytechnica 21 RESERCH RTICLE Calculation of the temperature of boundary

More information

RaneNote BESSEL FILTER CROSSOVER

RaneNote BESSEL FILTER CROSSOVER RaneNote BESSEL FILTER CROSSOVER A Beel Filter Croover, and It Relation to Other Croover Beel Function Phae Shift Group Delay Beel, 3dB Down Introduction One of the way that a croover may be contructed

More information

Layering as Optimization Decomposition

Layering as Optimization Decomposition TLEN7000/ECEN7002: Analytical Foundation of Network Layering a Optimization Decompoition Lijun Chen 11/29/2012 The Internet Compleity i ever increaing Large in ize and cope Enormou heterogeneity Incomplete

More information

THE IDENTIFICATION OF THE OPERATING REGIMES OF THE CONTROLLERS BY THE HELP OF THE PHASE TRAJECTORY

THE IDENTIFICATION OF THE OPERATING REGIMES OF THE CONTROLLERS BY THE HELP OF THE PHASE TRAJECTORY Mariu M. B LA Aurel Vlaicu Univerity of Arad, Engineering Faculty Bd. Revolu iei nr. 77, 3030, Arad, Romania, E-mail: mariu.bala@ieee.org THE IDENTIFICATION OF THE OPERATING REGIMES OF THE CONTROLLERS

More information

Design spacecraft external surfaces to ensure 95 percent probability of no mission-critical failures from particle impact.

Design spacecraft external surfaces to ensure 95 percent probability of no mission-critical failures from particle impact. PREFERRED RELIABILITY PAGE 1 OF 6 PRACTICES METEOROIDS & SPACE DEBRIS Practice: Deign pacecraft external urface to enure 95 percent probability of no miion-critical failure from particle impact. Benefit:

More information

Given the following circuit with unknown initial capacitor voltage v(0): X(s) Immediately, we know that the transfer function H(s) is

Given the following circuit with unknown initial capacitor voltage v(0): X(s) Immediately, we know that the transfer function H(s) is EE 4G Note: Chapter 6 Intructor: Cheung More about ZSR and ZIR. Finding unknown initial condition: Given the following circuit with unknown initial capacitor voltage v0: F v0/ / Input xt 0Ω Output yt -

More information

On the Robustness of the Characteristics Related to (M\M\1) ( \FCFS) Queue System Model

On the Robustness of the Characteristics Related to (M\M\1) ( \FCFS) Queue System Model www.ijemr.net ISSN (ONINE): 5-758, ISSN (PRINT): 394-696 Volume-5, Iue-3, June 5 International Journal of Engineering and Management Reearch Page Number: 839-847 On the Robutne of the haracteritic Related

More information

Online Parallel Scheduling of Non-uniform Tasks: Trading Failures for Energy

Online Parallel Scheduling of Non-uniform Tasks: Trading Failures for Energy Online Parallel Scheduling of Non-uniform Tak: Trading Failure for Energy Antonio Fernández Anta a, Chryi Georgiou b, Dariuz R. Kowalki c, Elli Zavou a,d,1 a Intitute IMDEA Network b Univerity of Cypru

More information

Pricing surplus server capacity for mean waiting time sensitive customers

Pricing surplus server capacity for mean waiting time sensitive customers Pricing urplu erver capacity for mean waiting time enitive cutomer Sudhir K. Sinha, N. Rangaraj and N. Hemachandra Indutrial Engineering and Operation Reearch, Indian Intitute of Technology Bombay, Mumbai

More information

On the Isomorphism of Fractional Factorial Designs 1

On the Isomorphism of Fractional Factorial Designs 1 journal of complexity 17, 8697 (2001) doi:10.1006jcom.2000.0569, available online at http:www.idealibrary.com on On the Iomorphim of Fractional Factorial Deign 1 Chang-Xing Ma Department of Statitic, Nankai

More information

[Saxena, 2(9): September, 2013] ISSN: Impact Factor: INTERNATIONAL JOURNAL OF ENGINEERING SCIENCES & RESEARCH TECHNOLOGY

[Saxena, 2(9): September, 2013] ISSN: Impact Factor: INTERNATIONAL JOURNAL OF ENGINEERING SCIENCES & RESEARCH TECHNOLOGY [Saena, (9): September, 0] ISSN: 77-9655 Impact Factor:.85 IJESRT INTERNATIONAL JOURNAL OF ENGINEERING SCIENCES & RESEARCH TECHNOLOGY Contant Stre Accelerated Life Teting Uing Rayleigh Geometric Proce

More information

Lecture 8: Period Finding: Simon s Problem over Z N

Lecture 8: Period Finding: Simon s Problem over Z N Quantum Computation (CMU 8-859BB, Fall 205) Lecture 8: Period Finding: Simon Problem over Z October 5, 205 Lecturer: John Wright Scribe: icola Rech Problem A mentioned previouly, period finding i a rephraing

More information

STOCHASTIC GENERALIZED TRANSPORTATION PROBLEM WITH DISCRETE DISTRIBUTION OF DEMAND

STOCHASTIC GENERALIZED TRANSPORTATION PROBLEM WITH DISCRETE DISTRIBUTION OF DEMAND OPERATIONS RESEARCH AND DECISIONS No. 4 203 DOI: 0.5277/ord30402 Marcin ANHOLCER STOCHASTIC GENERALIZED TRANSPORTATION PROBLEM WITH DISCRETE DISTRIBUTION OF DEMAND The generalized tranportation problem

More information

EE Control Systems LECTURE 14

EE Control Systems LECTURE 14 Updated: Tueday, March 3, 999 EE 434 - Control Sytem LECTURE 4 Copyright FL Lewi 999 All right reerved ROOT LOCUS DESIGN TECHNIQUE Suppoe the cloed-loop tranfer function depend on a deign parameter k We

More information

THE PARAMETERIZATION OF ALL TWO-DEGREES-OF-FREEDOM SEMISTRONGLY STABILIZING CONTROLLERS. Tatsuya Hoshikawa, Kou Yamada and Yuko Tatsumi

THE PARAMETERIZATION OF ALL TWO-DEGREES-OF-FREEDOM SEMISTRONGLY STABILIZING CONTROLLERS. Tatsuya Hoshikawa, Kou Yamada and Yuko Tatsumi International Journal of Innovative Computing, Information Control ICIC International c 206 ISSN 349-498 Volume 2, Number 2, April 206 pp. 357 370 THE PARAMETERIZATION OF ALL TWO-DEGREES-OF-FREEDOM SEMISTRONGLY

More information

Social Studies 201 Notes for March 18, 2005

Social Studies 201 Notes for March 18, 2005 1 Social Studie 201 Note for March 18, 2005 Etimation of a mean, mall ample ize Section 8.4, p. 501. When a reearcher ha only a mall ample ize available, the central limit theorem doe not apply to the

More information

LDPC Convolutional Codes Based on Permutation Polynomials over Integer Rings

LDPC Convolutional Codes Based on Permutation Polynomials over Integer Rings LDPC Convolutional Code Baed on Permutation Polynomial over Integer Ring Marco B. S. Tavare and Gerhard P. Fettwei Vodafone Chair Mobile Communication Sytem, Dreden Univerity of Technology, 01062 Dreden,

More information

Modeling The Interaction between TCP and Rate Adaptation at Links

Modeling The Interaction between TCP and Rate Adaptation at Links Modeling The Interaction between TCP and Rate Adaptation at Link Faheem Khuhawar, Marco Mellia, Michela Meo Dip. di Elettronica e Telecomunicazioni, Politecnico di Torino, Italy. Email: {firtname.latname}@polito.it

More information

Molecular Dynamics Simulations of Nonequilibrium Effects Associated with Thermally Activated Exothermic Reactions

Molecular Dynamics Simulations of Nonequilibrium Effects Associated with Thermally Activated Exothermic Reactions Original Paper orma, 5, 9 7, Molecular Dynamic Simulation of Nonequilibrium Effect ociated with Thermally ctivated Exothermic Reaction Jerzy GORECKI and Joanna Natalia GORECK Intitute of Phyical Chemitry,

More information

MATEMATIK Datum: Tid: eftermiddag. A.Heintz Telefonvakt: Anders Martinsson Tel.:

MATEMATIK Datum: Tid: eftermiddag. A.Heintz Telefonvakt: Anders Martinsson Tel.: MATEMATIK Datum: 20-08-25 Tid: eftermiddag GU, Chalmer Hjälpmedel: inga A.Heintz Telefonvakt: Ander Martinon Tel.: 073-07926. Löningar till tenta i ODE och matematik modellering, MMG5, MVE6. Define what

More information

Lecture 7: Testing Distributions

Lecture 7: Testing Distributions CSE 5: Sublinear (and Streaming) Algorithm Spring 014 Lecture 7: Teting Ditribution April 1, 014 Lecturer: Paul Beame Scribe: Paul Beame 1 Teting Uniformity of Ditribution We return today to property teting

More information

CHAPTER 8 OBSERVER BASED REDUCED ORDER CONTROLLER DESIGN FOR LARGE SCALE LINEAR DISCRETE-TIME CONTROL SYSTEMS

CHAPTER 8 OBSERVER BASED REDUCED ORDER CONTROLLER DESIGN FOR LARGE SCALE LINEAR DISCRETE-TIME CONTROL SYSTEMS CHAPTER 8 OBSERVER BASED REDUCED ORDER CONTROLLER DESIGN FOR LARGE SCALE LINEAR DISCRETE-TIME CONTROL SYSTEMS 8.1 INTRODUCTION 8.2 REDUCED ORDER MODEL DESIGN FOR LINEAR DISCRETE-TIME CONTROL SYSTEMS 8.3

More information

μ + = σ = D 4 σ = D 3 σ = σ = All units in parts (a) and (b) are in V. (1) x chart: Center = μ = 0.75 UCL =

μ + = σ = D 4 σ = D 3 σ = σ = All units in parts (a) and (b) are in V. (1) x chart: Center = μ = 0.75 UCL = Our online Tutor are available 4*7 to provide Help with Proce control ytem Homework/Aignment or a long term Graduate/Undergraduate Proce control ytem Project. Our Tutor being experienced and proficient

More information

Nonlinear Single-Particle Dynamics in High Energy Accelerators

Nonlinear Single-Particle Dynamics in High Energy Accelerators Nonlinear Single-Particle Dynamic in High Energy Accelerator Part 6: Canonical Perturbation Theory Nonlinear Single-Particle Dynamic in High Energy Accelerator Thi coure conit of eight lecture: 1. Introduction

More information

Jul 4, 2005 turbo_code_primer Revision 0.0. Turbo Code Primer

Jul 4, 2005 turbo_code_primer Revision 0.0. Turbo Code Primer Jul 4, 5 turbo_code_primer Reviion. Turbo Code Primer. Introduction Thi document give a quick tutorial on MAP baed turbo coder. Section develop the background theory. Section work through a imple numerical

More information

ied widely. The waiting time ditribution and queue ize for a queue with gamma ditributed arrival proce with a pecial Markovian tructure and i.i.d. exp

ied widely. The waiting time ditribution and queue ize for a queue with gamma ditributed arrival proce with a pecial Markovian tructure and i.i.d. exp Modeling of Correlated Arrival Procee in the Internet Muckai K Girih SBC Technology Reource, Inc. 4698 Willow Road Pleaanton, CA 94588 mgirih@tri.bc.com Jian-Qiang Hu Λ Dept. of Manufacturing Engineering

More information

Finite Element Analysis of a Fiber Bragg Grating Accelerometer for Performance Optimization

Finite Element Analysis of a Fiber Bragg Grating Accelerometer for Performance Optimization Finite Element Analyi of a Fiber Bragg Grating Accelerometer for Performance Optimization N. Baumallick*, P. Biwa, K. Dagupta and S. Bandyopadhyay Fiber Optic Laboratory, Central Gla and Ceramic Reearch

More information

A Bluffer s Guide to... Sphericity

A Bluffer s Guide to... Sphericity A Bluffer Guide to Sphericity Andy Field Univerity of Suex The ue of repeated meaure, where the ame ubject are teted under a number of condition, ha numerou practical and tatitical benefit. For one thing

More information

EFFECT ON PERSISTENCE OF INTRA-SPECIFIC COMPETITION IN COMPETITION MODELS

EFFECT ON PERSISTENCE OF INTRA-SPECIFIC COMPETITION IN COMPETITION MODELS Electronic Journal of Differential Equation, Vol. 2007(2007, No. 25, pp. 0. ISSN: 072-669. URL: http://ejde.math.txtate.edu or http://ejde.math.unt.edu ftp ejde.math.txtate.edu (login: ftp EFFECT ON PERSISTENCE

More information

Stochastic Neoclassical Growth Model

Stochastic Neoclassical Growth Model Stochatic Neoclaical Growth Model Michael Bar May 22, 28 Content Introduction 2 2 Stochatic NGM 2 3 Productivity Proce 4 3. Mean........................................ 5 3.2 Variance......................................

More information

Unavoidable Cycles in Polynomial-Based Time-Invariant LDPC Convolutional Codes

Unavoidable Cycles in Polynomial-Based Time-Invariant LDPC Convolutional Codes European Wirele, April 7-9,, Vienna, Autria ISBN 978--87-4-9 VE VERLAG GMBH Unavoidable Cycle in Polynomial-Baed Time-Invariant LPC Convolutional Code Hua Zhou and Norbert Goertz Intitute of Telecommunication

More information

Linear Quadratic Stochastic Differential Games under Asymmetric Value of Information

Linear Quadratic Stochastic Differential Games under Asymmetric Value of Information Preprint of the 2th World Congre The International Federation of Automatic Control Touloue, France, July 9-4, 27 Linear Quadratic Stochatic Differential Game under Aymmetric Value of Information Dipankar

More information

ISSN: ISO 9001:2008 Certified International Journal of Engineering and Innovative Technology (IJEIT) Volume 5, Issue 5, November 2015

ISSN: ISO 9001:2008 Certified International Journal of Engineering and Innovative Technology (IJEIT) Volume 5, Issue 5, November 2015 Optimum Deign of Charge Pump Circuit Uing Genetic Algorithm Dr. Ahmad T. Youni, Shamil H. Huein, and Ahmad A. Imail Abtract Integrated charge pump circuit are power converter ued to obtain a different

More information

Improving the Efficiency of a Digital Filtering Scheme for Diabatic Initialization

Improving the Efficiency of a Digital Filtering Scheme for Diabatic Initialization 1976 MONTHLY WEATHER REVIEW VOLUME 15 Improving the Efficiency of a Digital Filtering Scheme for Diabatic Initialization PETER LYNCH Met Éireann, Dublin, Ireland DOMINIQUE GIARD CNRM/GMAP, Météo-France,

More information

THE EXPERIMENTAL PERFORMANCE OF A NONLINEAR DYNAMIC VIBRATION ABSORBER

THE EXPERIMENTAL PERFORMANCE OF A NONLINEAR DYNAMIC VIBRATION ABSORBER Proceeding of IMAC XXXI Conference & Expoition on Structural Dynamic February -4 Garden Grove CA USA THE EXPERIMENTAL PERFORMANCE OF A NONLINEAR DYNAMIC VIBRATION ABSORBER Yung-Sheng Hu Neil S Ferguon

More information