Performance-Energy Trade-off in Multi-Server Queueing Systems with Setup Delay
|
|
- Karen Pearson
- 6 years ago
- Views:
Transcription
1 Performance-Energy Trade-off n Mlt-Server Qeeng Systems wth Setp Delay Saml Aalto Aalto Unversty, Fnland LCCC Clod Comptng Workshop 7 9 May 2014 Lnd, Sweden
2 Co-operaton wth Esa Hyytä, Pas Lassla, Mskr Gebrehwot (Aalto Unversty Rhonda Rghter (UC Berkeley 2
3 Qeeng models Sngle-server qee (M/G/1 Parallel qees l m homogeneos servers l m m Mlt-server qee (M/M/n l 1 m heterogeneos servers l m 1 n m m n Introdcton 3
4 Performance-energy trade-off Energy saved by swtchng the server off when dle However, performance mpared, f swtchng the server back on takes tme (setp delay BUSY / IDLE / OFF / SETUP? x x x x Introdcton 4
5 Cost model Performance: E[T] = mean delay per job (n seconds E[X] = mean nmber of jobs = le[t] Defnton: delay = response tme Energy: E[E] = mean energy per job (n joles E[P] = mean power consmed = le[e] Power consmpton levels: 0 P P P P off dle setp bsy Introdcton 5
6 Objectve fncton Energy-Response-tme- Weghted-Sm (ERWS: E[T] + E[E]/b e.g. Werman & al. (2009 Energy-Response-tme-Prodct (ERP: E[T]E[E] e.g. Gandh & al. (2010b General form: w 1 E[T] t1 E[E] e1 + w 2 E[T] t2 E[E] e2 by Macco & Down (2013 ERWS: w 1 = 1, t 1 = 1, e 1 = 0 w 2 = 1/b, t 2 = 0, e 2 = 1 ERP: w 1 = 1, t 1 = 1, e 1 = 1 w 2 = 0, t 2 = 0, e 2 = 0 Introdcton 6
7 Part I Sngle-server qee wth setp delays l m 7 Part I Sngle-server qee
8 Optmal swtchng on/off polcy Macco & Down (2013 M/G/1-FIFO Setp delay D generally dstrbted wth mean 1/g Control parameters: Delayed swtch-off for an exponental tme wth mean 1/a Server swtched on after k new job arrvals Objectve fncton: Gen. form [ ] 1 E[ ] 1 E[ ] 2 E[ ] 2 1 E t e t e T E w2 T E w + Polces: NEVEROFF: a 0 DELAYEDOFF: 0 a INSTANTOFF: a Theorem: For ERWS objectve fncton optmal polcy s ether NEVEROFF or INSTANTOFF Smlar reslt n Gandh & al. (2010b for ERP objectve fncton 8 Part I Sngle-server qee
9 Optmal swtchng on/off polcy Gebrehwot & al. (2014 M/G/1-FIFO Setp delay D generally dstrbted wth mean 1/g Control parameters: Delayed swtch-off for a gen. dstrbted tme wth mean 1/a Server swtched on after k new job arrvals Objectve fncton: Gen. form [ ] 1 E[ ] 1 E[ ] 2 E[ ] 2 1 E t e t e T E w2 T E w + Polces: NEVEROFF: a 0 DELAYEDOFF: 0 a INSTANTOFF: a Theorem: For gen. objectve fncton optmal polcy s ether NEVEROFF or INSTANTOFF NEVEROFF s better f P dle s sffcently small compared to P setp 9 Part I Sngle-server qee
10 Part II Mlt-server qee wth setp delays l 1 m n m 10 Part II Mlt-server qee
11 Analyss of server farms wth setp delays Gandh & al. (2010a M/M/n Setp delay D exponentally dstrbted Objectve fncton: Separately E[T] and E[P] Polces: ON/IDLE = NEVEROFF ON/OFF = INSTANTOFF ON/OFF/STAG = INSTANTOFF wth staggered bootp Mxed polcy: ON/IDLE(t swtchng dle server off only f nr of bsy and dle servers > t Conclsons: Under hgh loads, trnng servers off can reslt n hgher power consmpton and far hgher response tmes. As the sze of the server farm s ncreased, the advantages of trnng servers off ncrease. 11 Part II Mlt-server qee
12 Analyss of server farms wth setp delays Gandh & al. (2010a 12 Part II Mlt-server qee
13 Optmzaton of server farms wth setp delay Gandh & al. (2010b M/M/n Setp delay determnstc Addtonal sleep states S wth 0 P P P off and determnstc (setp delays 0 d d d dle sleep sleep Objectve fncton: ERP E[ T ] E[ P] dle off Polces: NEVEROFF INSTANTOFF SLEEP(S Probablstc and other Theorem: For n = 1, optmal statc control s ether NEVEROFF, INSTANTOFF or SLEEP(S Robst polcy: DELAYEDOFF wth MRB (Most Recent Bsy 13 Part II Mlt-server qee
14 Optmzaton of server farms wth setp delay Gandh & al. (2010b 14 Part II Mlt-server qee
15 Part III Parallel qees wth setp delays l m 1 m n 15 Part III Parallel qees
16 Dspatchng problem Dspatchng = Task assgnment = Rotng Random job arrvals wth random servce reqrements Dspatchng decson made pon the arrval l m 1 m n Or settng: M/G/. Posson arrvals generally dstrbted job szes heterogeneos servers wth FIFO qeeng dscplne (NEVEROFF or INSTANTOFF 16 Part III Parallel qees
17 Statc dspatchng polces RND = Bernoll splttng choose the qee pre randomly no sze nor state nformaton needed SITA = Sze Interval Task Assgnment choose the qee wth smlar jobs based on the sze of the arrvng job, bt no state nformaton needed Harchol-Balter et al. ( Part III Parallel qees
18 MDP approach Any statc polcy (RND, SITA reslts n parallel M/G/1 qees Fx the statc polcy and determne relatve vales for all these parallel M/G/1 qees Dspatch the arrvng job to the qee that mnmzes the mean addtonal costs As the reslt, yo get a better dynamc dspatchng polcy Ths s called Frst Polcy Iteraton (FPI n the MDP theory Applcable for the ERWS objectve fncton 18 Part III Parallel qees
19 Relatve vales Defnton: For a fxed polcy resltng n a stable system, the vale fncton v(x gves the expected dfference n the nfnte horzon cmlatve costs between the system ntally n state x, and the system ntally n eqlbrm Defnton: For a fxed polcy resltng n a stable system, the relatve vale v(x - v(0 gves the expected dfference n the nfnte horzon cmlatve costs between the system ntally n state x, and the system ntally n state 0 19 Part III Parallel qees
20 Sze-aware M/G/1 qee wthot setp delays Hyytä et al. (2012 State descrpton: D1 + + D n D = remanng servce tme of job = backlog = nfnshed work Mean vales: Reslt: Sze-aware relatve vales v v T P E[ T ] E[ P] ( - v ( - v E[ S] + (1 - P T P (0 (0 le[ S 2 ] 2(1- dle l 2 2(1- ( P + P bsy - bsy P dle 20 Part III Parallel qees
21 Sze-aware M/G/1 qee wth setp delays Hyytä et al. (2014a State descrpton: D0 + D1 + + D n D = remanng servce tme of job D 0 = remanng setp delay = vrtal backlog Assme: Determnstc setp delay d and Psetp P bsy Mean vales: Reslt: Sze-aware relatve vales v v T P E[ T ] E[ P] ( - v ( - v E[ S] + + ld 1+ ld T P (0 (0 le[ S 2 ] 2(1- P bsy l 2 2(1-1+ ld - P + d (2+ ld 2(1+ ld ld (2+ ld 2(1- (1+ ld bsy 21 Part III Parallel qees
22 FPI polcy Hyytä et al. (2012, 2014a For NEVEROFF servers: Dspatch the job wth servce tme x to qee mnmzng the mean addtonal costs: For INSTANTOFF servers: Dspatch the job wth servce tme x to qee mnmzng the mean addtonal costs: 22 Part III Parallel qees, ( 0, 1( (,, (, ( 0, 1( ( 0 1(,, ( v d x v x a v d x v d x x a P P P T T T , (, (,, (, (, (,, ( v x v x a v x v x x a P P P T T T
23 Nmercal reslts Hyytä et al. (2014a 23 Part III Parallel qees
24 Nmercal reslts Hyytä et al. (2014a 24 Part III Parallel qees
25 Nmercal reslts Hyytä et al. (2014a 25 Part III Parallel qees
26 Other qeeng dscplnes Hyytä et al. (2014b LIFO n the M/G/. settng wth setp delays PS n the M/D/. settng wth setp delays Bt t s another story 26 Part III Parallel qees
27 References Harchol-Balter, Crovella & Mrta (1999 On Choosng a Task Assgnment Polcy for a Dstrbted Server System, JPDC Werman, Andrew & Tang (2009 Power-aware speed scalng n processor sharng systems, n IEEE INFOCOM Gandh, Harchol-Balter & Adan (2010a Server farms wth setp costs, PEVA Gandh, Gpta, Harchol-Balter & Kozch (2010b Optmalty analyss of energy-performance trade-off for server farm management, PEVA Hyytä, Penttnen & Aalto (2012 Sze- and state-aware dspatchng problem wth qee-specfc job szes, EJOR Macco & Down (2013 On optmal polces for energy-aware servers, n MASCOTS Hyytä, Rghter & Aalto (2014a Task assgnment n a heterogeneos server farm wth swtchng delays and general energy-aware cost strctre, to appear n PEVA Hyytä, Rghter & Aalto (2014b Energy-aware job assgnment n server farms wth setp delays nder LCFS and PS, accepted to ITC Gebrehwot, Lassla & Aalto (2014 Energy-aware qeeng models and controls for server farms, ongong work 27 References
28 The End 28
State-dependent and Energy-aware Control of Server Farm
State-dependent and Energy-aware Control of Server Farm Esa Hyytiä, Rhonda Righter and Samuli Aalto Aalto University, Finland UC Berkeley, USA First European Conference on Queueing Theory ECQT 2014 First
More informationMinimizing Slowdown in Heterogeneous Size-Aware Dispatching Systems (full version)
Mnmzng Slowdown n Heterogeneous Sze-Aware Dspatchng Systems (full verson) Esa Hyytä, Samul Aalto and Aleks Penttnen Department of Communcatons and Networkng Aalto Unversty School of Electrcal Engneerng,
More informationDynamic Control of Running Servers
Dynamic Control of Running Servers Esa Hyytiä 1,2, Douglas Down 3, Pasi Lassila 2, and Samuli Aalto 2 1 Department of Computer Science, University of Iceland, Iceland 2 Department of Communications and
More informationTask Assignment in a Heterogeneous Server Farm with Switching Delays and General Energy-Aware Cost Structure
Task Assignment in a Heterogeneous Server Farm with Switching Delays and General Energy-Aware Cost Structure Esa Hyytiä a,,, Rhonda Righter b, Samuli Aalto a a Department of Communications and Networking,
More informationMinimisation of the Average Response Time in a Cluster of Servers
Mnmsaton of the Average Response Tme n a Cluster of Servers Valery Naumov Abstract: In ths paper, we consder task assgnment problem n a cluster of servers. We show that optmal statc task assgnment s tantamount
More informationQueueing Networks II Network Performance
Queueng Networks II Network Performance Davd Tpper Assocate Professor Graduate Telecommuncatons and Networkng Program Unversty of Pttsburgh Sldes 6 Networks of Queues Many communcaton systems must be modeled
More informationReal-Time Systems. Multiprocessor scheduling. Multiprocessor scheduling. Multiprocessor scheduling
Real-Tme Systems Multprocessor schedulng Specfcaton Implementaton Verfcaton Multprocessor schedulng -- -- Global schedulng How are tasks assgned to processors? Statc assgnment The processor(s) used for
More informationLecture 4: November 17, Part 1 Single Buffer Management
Lecturer: Ad Rosén Algorthms for the anagement of Networs Fall 2003-2004 Lecture 4: November 7, 2003 Scrbe: Guy Grebla Part Sngle Buffer anagement In the prevous lecture we taled about the Combned Input
More informationIntroduction to Continuous-Time Markov Chains and Queueing Theory
Introducton to Contnuous-Tme Markov Chans and Queueng Theory From DTMC to CTMC p p 1 p 12 1 2 k-1 k p k-1,k p k-1,k k+1 p 1 p 21 p k,k-1 p k,k-1 DTMC 1. Transtons at dscrete tme steps n=,1,2, 2. Past doesn
More informationECE697AA Lecture 17. Birth-Death Processes
Tlman Wolf Department of Electrcal and Computer Engneerng /4/8 ECE697AA ecture 7 Queung Systems II ECE697AA /4/8 Uass Amherst Tlman Wolf Brth-Death Processes Solvng general arov chan can be dffcult Smpler,
More informationEnsemble Methods: Boosting
Ensemble Methods: Boostng Ncholas Ruozz Unversty of Texas at Dallas Based on the sldes of Vbhav Gogate and Rob Schapre Last Tme Varance reducton va baggng Generate new tranng data sets by samplng wth replacement
More informationEquilibrium Analysis of the M/G/1 Queue
Eulbrum nalyss of the M/G/ Queue Copyrght, Sanay K. ose. Mean nalyss usng Resdual Lfe rguments Secton 3.. nalyss usng an Imbedded Marov Chan pproach Secton 3. 3. Method of Supplementary Varables done later!
More informationEmbedded Systems. 4. Aperiodic and Periodic Tasks
Embedded Systems 4. Aperodc and Perodc Tasks Lothar Thele 4-1 Contents of Course 1. Embedded Systems Introducton 2. Software Introducton 7. System Components 10. Models 3. Real-Tme Models 4. Perodc/Aperodc
More informationLecture 14: Bandits with Budget Constraints
IEOR 8100-001: Learnng and Optmzaton for Sequental Decson Makng 03/07/16 Lecture 14: andts wth udget Constrants Instructor: Shpra Agrawal Scrbed by: Zhpeng Lu 1 Problem defnton In the regular Mult-armed
More informationThe Bellman Equation
The Bellman Eqaton Reza Shadmehr In ths docment I wll rovde an elanaton of the Bellman eqaton, whch s a method for otmzng a cost fncton and arrvng at a control olcy.. Eamle of a game Sose that or states
More informationAssignment 5. Simulation for Logistics. Monti, N.E. Yunita, T.
Assgnment 5 Smulaton for Logstcs Mont, N.E. Yunta, T. November 26, 2007 1. Smulaton Desgn The frst objectve of ths assgnment s to derve a 90% two-sded Confdence Interval (CI) for the average watng tme
More informationReading Assignment. Panel Data Cross-Sectional Time-Series Data. Chapter 16. Kennedy: Chapter 18. AREC-ECON 535 Lec H 1
Readng Assgnment Panel Data Cross-Sectonal me-seres Data Chapter 6 Kennedy: Chapter 8 AREC-ECO 535 Lec H Generally, a mxtre of cross-sectonal and tme seres data y t = β + β x t + β x t + + β k x kt + e
More informationCHAPTER 17 Amortized Analysis
CHAPTER 7 Amortzed Analyss In an amortzed analyss, the tme requred to perform a sequence of data structure operatons s averaged over all the operatons performed. It can be used to show that the average
More informationCoarse-Grain MTCMOS Sleep
Coarse-Gran MTCMOS Sleep Transstor Szng Usng Delay Budgetng Ehsan Pakbazna and Massoud Pedram Unversty of Southern Calforna Dept. of Electrcal Engneerng DATE-08 Munch, Germany Leakage n CMOS Technology
More informationOrder Full Rate, Leadtime Variability, and Advance Demand Information in an Assemble- To-Order System
Order Full Rate, Leadtme Varablty, and Advance Demand Informaton n an Assemble- To-Order System by Lu, Song, and Yao (2002) Presented by Png Xu Ths summary presentaton s based on: Lu, Yngdong, and Jng-Sheng
More informationVariability-Driven Module Selection with Joint Design Time Optimization and Post-Silicon Tuning
Asa and South Pacfc Desgn Automaton Conference 2008 Varablty-Drven Module Selecton wth Jont Desgn Tme Optmzaton and Post-Slcon Tunng Feng Wang, Xaoxa Wu, Yuan Xe The Pennsylvana State Unversty Department
More informationLast Time. Priority-based scheduling. Schedulable utilization Rate monotonic rule: Keep utilization below 69% Static priorities Dynamic priorities
Last Tme Prorty-based schedulng Statc prortes Dynamc prortes Schedulable utlzaton Rate monotonc rule: Keep utlzaton below 69% Today Response tme analyss Blockng terms Prorty nverson And solutons Release
More informationGPU friendly Fast Poisson Solver for Structured Power Grid Network Analysis
GPU frendly Fast Posson Solver for Strctred Power Grd Network Analyss Jn Sh, Yc Ca, Xaoy Wang Dept of Compter Scence and Technology, Tsngha Unv. Wentng Ho, Lwe Ma, Pe-Hsn Ho Synopsys Inc. Sheldon X.-D.
More informationOverhead-Aware Compositional Analysis of Real-Time Systems
Overhead-Aware ompostonal Analyss of Real-Tme Systems Lnh T.X. Phan, Meng Xu, Jaewoo Lee, nsup Lee, Oleg Sokolsky PRESE enter Department of omputer and nformaton Scence Unversty of Pennsylvana ompostonal
More informationSingle-Facility Scheduling over Long Time Horizons by Logic-based Benders Decomposition
Sngle-Faclty Schedulng over Long Tme Horzons by Logc-based Benders Decomposton Elvn Coban and J. N. Hooker Tepper School of Busness, Carnege Mellon Unversty ecoban@andrew.cmu.edu, john@hooker.tepper.cmu.edu
More informationOn the Interaction between Load Balancing and Speed Scaling
IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. XX, NO. XX, DECEMBER 05 On the Interacton between Load Balancng and Speed Scalng Ljun Chen, Member, IEEE, and Na L, Member, IEEE Abstract Speed scalng
More informationCorrelation Clustering with Noisy Input
Correlaton Clsterng wth Nosy Inpt Clare Mathe Warren Schdy Brown Unversty SODA 2010 Nosy Correlaton Clsterng Model Unknown base clsterng B of n obects Nose: each edge s controlled by an adversary wth probablty
More informationAnalysis of Queuing Delay in Multimedia Gateway Call Routing
Analyss of Queung Delay n Multmeda ateway Call Routng Qwe Huang UTtarcom Inc, 33 Wood Ave. outh Iseln, NJ 08830, U..A Errol Lloyd Computer Informaton cences Department, Unv. of Delaware, Newark, DE 976,
More informationCLOSED-FORM CHARACTERIZATION OF THE CHANNEL CAPACITY OF MULTI-BRANCH MAXIMAL RATIO COMBINING OVER CORRELATED NAKAGAMI FADING CHANNELS
CLOSED-FORM CHARACTERIZATION OF THE CHANNEL CAPACITY OF MULTI-BRANCH MAXIMAL RATIO COMBINING OVER CORRELATED NAKAGAMI FADING CHANNELS Yawgeng A. Cha and Karl Yng-Ta Hang Department of Commncaton Engneerng,
More informationECE559VV Project Report
ECE559VV Project Report (Supplementary Notes Loc Xuan Bu I. MAX SUM-RATE SCHEDULING: THE UPLINK CASE We have seen (n the presentaton that, for downlnk (broadcast channels, the strategy maxmzng the sum-rate
More informationSimultaneous Optimization of Berth Allocation, Quay Crane Assignment and Quay Crane Scheduling Problems in Container Terminals
Smultaneous Optmzaton of Berth Allocaton, Quay Crane Assgnment and Quay Crane Schedulng Problems n Contaner Termnals Necat Aras, Yavuz Türkoğulları, Z. Caner Taşkın, Kuban Altınel Abstract In ths work,
More informationOpen and Closed Networks of M/M/m Type Queues (Jackson s Theorem for Open and Closed Networks) Copyright 2015, Sanjay K. Bose 1
Ope ad Closed Networks of //m Type Qees Jackso s Theorem for Ope ad Closed Networks Copyrght 05, Saay. Bose p osso Rate λp osso rocess Average Rate λ p osso Rate λp N p p N osso Rate λp N Splttg a osso
More informationOutline. Communication. Bellman Ford Algorithm. Bellman Ford Example. Bellman Ford Shortest Path [1]
DYNAMIC SHORTEST PATH SEARCH AND SYNCHRONIZED TASK SWITCHING Jay Wagenpfel, Adran Trachte 2 Outlne Shortest Communcaton Path Searchng Bellmann Ford algorthm Algorthm for dynamc case Modfcatons to our algorthm
More informationEffective Power Optimization combining Placement, Sizing, and Multi-Vt techniques
Effectve Power Optmzaton combnng Placement, Szng, and Mult-Vt technques Tao Luo, Davd Newmark*, and Davd Z Pan Department of Electrcal and Computer Engneerng, Unversty of Texas at Austn *Advanced Mcro
More informationIntroduction to Turbulence Modelling
Introdcton to Trblence Modellng 1 Nmercal methods 0 1 t Mathematcal descrpton p F Reslts For eample speed, pressre, temperatre Geometry Models for trblence, combston etc. Mathematcal descrpton of physcal
More informationMultiuser Cognitive Access of Continuous Time Markov Channels: Maximum Throughput and Effective Bandwidth Regions
Multuser Cogntve Access of Contnuous Tme Markov Channels: Maxmum Throughput and Effectve Bandwdth Regons Shyao Chen, and Lang Tong School of Electrcal and Computer Engneerng Cornell Unversty, Ithaca, NY
More informationA FAST HEURISTIC FOR TASKS ASSIGNMENT IN MANYCORE SYSTEMS WITH VOLTAGE-FREQUENCY ISLANDS
Shervn Haamn A FAST HEURISTIC FOR TASKS ASSIGNMENT IN MANYCORE SYSTEMS WITH VOLTAGE-FREQUENCY ISLANDS INTRODUCTION Increasng computatons n applcatons has led to faster processng. o Use more cores n a chp
More informationNetwork of Markovian Queues. Lecture
etwork of Markovan Queues etwork of Markovan Queues ETW09 20 etwork queue ed, G E ETW09 20 λ If the frst queue was not empty Then the tme tll the next arrval to the second queue wll be equal to the servce
More informationC e n t r u m v o o r W i s k u n d e e n I n f o r m a t i c a
C e n t r u m v o o r W s k u n d e e n I n f o r m a t c a Probablty, Networks and Algorthms Probablty, Networks and Algorthms Traffc-splttng networks operatng under alpha-far sharng polces and balanced
More informationAppendix B: Resampling Algorithms
407 Appendx B: Resamplng Algorthms A common problem of all partcle flters s the degeneracy of weghts, whch conssts of the unbounded ncrease of the varance of the mportance weghts ω [ ] of the partcles
More informationEEL 6266 Power System Operation and Control. Chapter 3 Economic Dispatch Using Dynamic Programming
EEL 6266 Power System Operaton and Control Chapter 3 Economc Dspatch Usng Dynamc Programmng Pecewse Lnear Cost Functons Common practce many utltes prefer to represent ther generator cost functons as sngle-
More informationCIE4801 Transportation and spatial modelling Trip distribution
CIE4801 ransportaton and spatal modellng rp dstrbuton Rob van Nes, ransport & Plannng 17/4/13 Delft Unversty of echnology Challenge the future Content What s t about hree methods Wth specal attenton for
More informationRevenue comparison when the length of horizon is known in advance. The standard errors of all numbers are less than 0.1%.
Golrezae, Nazerzadeh, and Rusmevchentong: Real-tme Optmzaton of Personalzed Assortments Management Scence 00(0, pp. 000 000, c 0000 INFORMS 37 Onlne Appendx Appendx A: Numercal Experments: Appendx to Secton
More informationStochastic Optimal Controls for Parallel-Server Channels with Zero Waiting Buffer Capacity and Multi-Class Customers
Advances n Systems Scence and Applcatons (009), Vol.9, No.4 67-646 Stochastc Optmal Controls for Parallel-Server Channels wth Zero Watng Buffer Capacty and Mult-Class Customers Wanyang Da and Youy Feng
More informationAN IMPROVED METHOD OF HIERARCHIC ANALYSIS FOR CHOOSING OPTIMAL INFORMATION PROTECTION SYSTEM IN COMPUTER NETWORKS
S. M. Kusemko, Cand. Sc. (Eng.), Ass. Prof.; V. M. Melnchuk AN IMPROVED METHOD OF HIERARCHIC ANALYSIS FOR CHOOSING OPTIMAL INFORMATION PROTECTION SYSTEM IN COMPUTER NETWORKS An mproved herarchc analyss
More informationESS 265 Spring Quarter 2005 Time Series Analysis: Error Analysis
ESS 65 Sprng Qarter 005 Tme Seres Analyss: Error Analyss Lectre 9 May 3, 005 Some omenclatre Systematc errors Reprodcbly errors that reslt from calbraton errors or bas on the part of the obserer. Sometmes
More informationAmiri s Supply Chain Model. System Engineering b Department of Mathematics and Statistics c Odette School of Business
Amr s Supply Chan Model by S. Ashtab a,, R.J. Caron b E. Selvarajah c a Department of Industral Manufacturng System Engneerng b Department of Mathematcs Statstcs c Odette School of Busness Unversty of
More informationPricing and Resource Allocation Game Theoretic Models
Prcng and Resource Allocaton Game Theoretc Models Zhy Huang Changbn Lu Q Zhang Computer and Informaton Scence December 8, 2009 Z. Huang, C. Lu, and Q. Zhang (CIS) Game Theoretc Models December 8, 2009
More informationSelf-Adaptive Admission Control Policies for Resource-Sharing Systems
Self-Adaptve Admsson Control Polces for Resource-Sharng Systems Varun Gupta Carnege Mellon Unversty varun@cs.cmu.edu Mor Harchol-Balter Carnege Mellon Unversty harchol@cs.cmu.edu ABSTRACT We consder the
More informationt } = Number of calls in progress at time t. Engsett Model (Erlang-B)
Engsett Model (Erlang-B) A B Desrpton: Bloed-alls lost model Consder a entral exhange wth users (susrers) sharng truns (truns). When >, long ours. Ths s the ase of prnpal nterest. Assume that the truns
More informationEFFECTS OF JOINT REPLENISHMENT POLICY ON COMPANY COST UNDER PERMISSIBLE DELAY IN PAYMENTS
Mathematcal and Computatonal Applcatons, Vol. 5, No., pp. 8-58,. Assocaton for Scentfc Research EFFECS OF JOIN REPLENISHMEN POLICY ON COMPANY COS UNDER PERMISSIBLE DELAY IN PAYMENS Yu-Chung sao, Mng-Yu
More informationEURANDOM PREPRINT SERIES Server farms with setup costs. A. Gandhi, M. Harchol-Balter, I. Adan ISSN
EURANDOM PREPRINT SERIES 200-07 Server farms with setup costs A. Gandhi, M. Harchol-Balter, I. Adan ISSN 389-2355 Server farms with setup costs Anshul Gandhi Mor Harchol-Balter Ivo Adan anshulg,harchol
More informationNonlinear Programming Formulations for On-line Applications
Nonlnear Programmng Formlatons for On-lne Applcatons L. T. Begler Carnege Mellon Unversty Janary 2007 Jont work wth Vctor Zavala and Carl Lard NLP for On-lne Process Control Nonlnear Model Predctve Control
More informationMeenu Gupta, Man Singh & Deepak Gupta
IJS, Vol., o. 3-4, (July-December 0, pp. 489-497 Serals Publcatons ISS: 097-754X THE STEADY-STATE SOLUTIOS OF ULTIPLE PARALLEL CHAELS I SERIES AD O-SERIAL ULTIPLE PARALLEL CHAELS BOTH WITH BALKIG & REEGIG
More informationThe Folded Normal Stochastic Frontier. Gholamreza Hajargasht Department of Economics University of Melbourne, Australia
The Folded Normal Stochastc Fronter Gholamreza Hajargasht Department of Economcs Unversty of Melborne, Astrala Abstract We ntrodce a stochastc fronter model wth a folded normal neffcency dstrbton. Ths
More informationCSC 411 / CSC D11 / CSC C11
18 Boostng s a general strategy for learnng classfers by combnng smpler ones. The dea of boostng s to take a weak classfer that s, any classfer that wll do at least slghtly better than chance and use t
More informationMETHOD OF MONOTONE STRUCTURAL EVOLUTION FOR CONTROL AND STATE CONSTRAINED OPTIMAL CONTROL PROBLEMS
METHOD OF MONOTONE STRUCTURAL EVOLUTION FOR CONTROL AND STATE CONSTRAINED OPTIMAL CONTROL PROBLEMS Macej Szymkat, Adam Korytowsk AGH Unversty of Scence and Technology -59 Kraków, Poland, e-mal: msz@a.agh.ed.pl
More informationToward Understanding Heterogeneity in Computing
Toward Understandng Heterogenety n Computng Arnold L. Rosenberg Ron C. Chang Department of Electrcal and Computer Engneerng Colorado State Unversty Fort Collns, CO, USA {rsnbrg, ron.chang@colostate.edu}
More informationChapter 3 Describing Data Using Numerical Measures
Chapter 3 Student Lecture Notes 3-1 Chapter 3 Descrbng Data Usng Numercal Measures Fall 2006 Fundamentals of Busness Statstcs 1 Chapter Goals To establsh the usefulness of summary measures of data. The
More informationAE/ME 339. K. M. Isaac. 8/31/2004 topic4: Implicit method, Stability, ADI method. Computational Fluid Dynamics (AE/ME 339) MAEEM Dept.
AE/ME 339 Comptatonal Fld Dynamcs (CFD) Comptatonal Fld Dynamcs (AE/ME 339) Implct form of dfference eqaton In the prevos explct method, the solton at tme level n,,n, depended only on the known vales of,
More informationA Make-to-Stock System with Multiple Customer Classes and Batch Ordering
OPERATIONS RESEARCH Vol. 56, No. 5, September October 28, pp. 1312 132 ssn 3-364X essn 1526-5463 8 565 1312 nforms do 1.1287/opre.18.549 28 INFORMS TECHNICAL NOTE A Make-to-Stock System wth Multple Customer
More informationA Simple Inventory System
A Smple Inventory System Lawrence M. Leems and Stephen K. Park, Dscrete-Event Smulaton: A Frst Course, Prentce Hall, 2006 Hu Chen Computer Scence Vrgna State Unversty Petersburg, Vrgna February 8, 2017
More informationRelaxations of Weakly Coupled Stochastic Dynamic Programs
OPERATIONS RESEARCH Vol. 56, No. 3, May June 2008, pp. 712 727 ssn 0030-364X essn 1526-5463 08 5603 0712 nforms do 10.1287/opre.1070.0445 2008 INFORMS Relaxatons of Weakly Coupled Stochastc Dynamc Programs
More informationCOMPLETE BUFFER SHARING IN ATM NETWORKS UNDER BURSTY ARRIVALS
COMPLETE BUFFER SHARING WITH PUSHOUT THRESHOLDS IN ATM NETWORKS UNDER BURSTY ARRIVALS Ozgur Aras and Tugrul Dayar Abstract. Broadband Integrated Servces Dgtal Networks (B{ISDNs) are to support multple
More informationCS 798: Homework Assignment 2 (Probability)
0 Sample space Assgned: September 30, 2009 In the IEEE 802 protocol, the congeston wndow (CW) parameter s used as follows: ntally, a termnal wats for a random tme perod (called backoff) chosen n the range
More informationPortfolios with Trading Constraints and Payout Restrictions
Portfolos wth Tradng Constrants and Payout Restrctons John R. Brge Northwestern Unversty (ont wor wth Chrs Donohue Xaodong Xu and Gongyun Zhao) 1 General Problem (Very) long-term nvestor (eample: unversty
More informationModule 9. Lecture 6. Duality in Assignment Problems
Module 9 1 Lecture 6 Dualty n Assgnment Problems In ths lecture we attempt to answer few other mportant questons posed n earler lecture for (AP) and see how some of them can be explaned through the concept
More informationApplication of Nonbinary LDPC Codes for Communication over Fading Channels Using Higher Order Modulations
Applcaton of Nonbnary LDPC Codes for Communcaton over Fadng Channels Usng Hgher Order Modulatons Rong-Hu Peng and Rong-Rong Chen Department of Electrcal and Computer Engneerng Unversty of Utah Ths work
More informationModeling Mood Variation and Covariation among Adolescent Smokers: Application of a Bivariate Location-Scale Mixed-Effects Model
Modelng Mood Varaton and Covaraton among Adolescent Smokers: Applcaton of a Bvarate Locaton-Scale Mxed-Effects Model Oksana Pgach, PhD, Donald Hedeker, PhD, Robn Mermelsten, PhD Insttte for Health Research
More informationCS : Algorithms and Uncertainty Lecture 17 Date: October 26, 2016
CS 29-128: Algorthms and Uncertanty Lecture 17 Date: October 26, 2016 Instructor: Nkhl Bansal Scrbe: Mchael Denns 1 Introducton In ths lecture we wll be lookng nto the secretary problem, and an nterestng
More informationTOPICS MULTIPLIERLESS FILTER DESIGN ELEMENTARY SCHOOL ALGORITHM MULTIPLICATION
1 2 MULTIPLIERLESS FILTER DESIGN Realzaton of flters wthout full-fledged multplers Some sldes based on support materal by W. Wolf for hs book Modern VLSI Desgn, 3 rd edton. Partly based on followng papers:
More informationWhat would be a reasonable choice of the quantization step Δ?
CE 108 HOMEWORK 4 EXERCISE 1. Suppose you are samplng the output of a sensor at 10 KHz and quantze t wth a unform quantzer at 10 ts per sample. Assume that the margnal pdf of the sgnal s Gaussan wth mean
More informationA Modeling System to Combine Optimization and Constraint. Programming. INFORMS, November Carnegie Mellon University.
A Modelng Sstem to Combne Optmzaton and Constrant Programmng John Hooker Carnege Mellon Unverst INFORMS November 000 Based on ont work wth Ignaco Grossmann Hak-Jn Km Mara Axlo Osoro Greger Ottosson Erlendr
More informationAsynchronous CSMA Policies in Multihop Wireless Networks with Primary Interference Constraints
Asynchronous CSMA Polces n Multhop Wreless Networks wth Prmary Interference Constrants Peter Marbach, Atlla Erylmaz, and Asu Ozdaglar Abstract We analyze Asynchronous Carrer Sense Multple Access (CSMA)
More informationOPTIMAL QUEUE-SIZE SCALING IN SWITCHED NETWORKS
The Annals of Appled Probablty 2014, Vol. 24, No. 6, 2207 2245 DOI: 10.1214/13-AAP970 Insttute of Mathematcal Statstcs, 2014 OPTIMAL QUEUE-SIZE SCALING IN SWITCHED NETWORKS BY D. SHAH 1,N.S.WALTON AND
More informationEngineering Risk Benefit Analysis
Engneerng Rsk Beneft Analyss.55, 2.943, 3.577, 6.938, 0.86, 3.62, 6.862, 22.82, ESD.72, ESD.72 RPRA 2. Elements of Probablty Theory George E. Apostolaks Massachusetts Insttute of Technology Sprng 2007
More informationThis paper is inspired by the recurring mismatch between demand and supply in the U.S. influenza vaccine
MANUFACTURING & SERVICE OPERATIONS MANAGEMENT Vol. 11, No. 4, Fall 2009, pp. 563 576 ssn 1523-4614 essn 1526-5498 09 1104 0563 nforms do 10.1287/msom.1080.0242 2009 INFORMS Cornot Competton Under Yeld
More informationProblem Set 1 Issued: Wednesday, February 11, 2015 Due: Monday, February 23, 2015
MASSACHUSETTS INSTITUTE OF TECHNOLOGY DEPARTMENT OF MECHANICAL ENGINEERING CAMBRIDGE MASSACHUSETTS 09.9 NUMERICAL FLUID MECHANICS SPRING 05 Problem Set Issed: Wednesday Febrary 05 De: Monday Febrary 05
More informationIntroduction to the R Statistical Computing Environment R Programming
Introducton to the R Statstcal Computng Envronment R Programmng John Fox McMaster Unversty ICPSR 2018 John Fox (McMaster Unversty) R Programmng ICPSR 2018 1 / 14 Programmng Bascs Topcs Functon defnton
More informationDynamic Programming. Lecture 13 (5/31/2017)
Dynamc Programmng Lecture 13 (5/31/2017) - A Forest Thnnng Example - Projected yeld (m3/ha) at age 20 as functon of acton taken at age 10 Age 10 Begnnng Volume Resdual Ten-year Volume volume thnned volume
More informationA LINEAR PROGRAM TO COMPARE MULTIPLE GROSS CREDIT LOSS FORECASTS. Dr. Derald E. Wentzien, Wesley College, (302) ,
A LINEAR PROGRAM TO COMPARE MULTIPLE GROSS CREDIT LOSS FORECASTS Dr. Derald E. Wentzen, Wesley College, (302) 736-2574, wentzde@wesley.edu ABSTRACT A lnear programmng model s developed and used to compare
More informationPresenters. Muscle Modeling. John Rasmussen (Presenter) Arne Kiis (Host) The web cast will begin in a few minutes.
Muscle Modelng If a part of my screen n mssng from your Vew, please press Sharng -> Vew -> Autoft The web cast wll begn n a few mnutes. Introducton (~5 mn) Overvew (~5 mn) Muscle knematcs (~10 mn) Muscle
More informationWhich Separator? Spring 1
Whch Separator? 6.034 - Sprng 1 Whch Separator? Mamze the margn to closest ponts 6.034 - Sprng Whch Separator? Mamze the margn to closest ponts 6.034 - Sprng 3 Margn of a pont " # y (w $ + b) proportonal
More informationOptimization Methods for Engineering Design. Logic-Based. John Hooker. Turkish Operational Research Society. Carnegie Mellon University
Logc-Based Optmzaton Methods for Engneerng Desgn John Hooker Carnege Mellon Unerst Turksh Operatonal Research Socet Ankara June 1999 Jont work wth: Srnas Bollapragada General Electrc R&D Omar Ghattas Cl
More informationDr. Shalabh Department of Mathematics and Statistics Indian Institute of Technology Kanpur
Analyss of Varance and Desgn of Exerments-I MODULE III LECTURE - 2 EXPERIMENTAL DESIGN MODELS Dr. Shalabh Deartment of Mathematcs and Statstcs Indan Insttute of Technology Kanur 2 We consder the models
More informationResource Allocation with a Budget Constraint for Computing Independent Tasks in the Cloud
Resource Allocaton wth a Budget Constrant for Computng Independent Tasks n the Cloud Wemng Sh and Bo Hong School of Electrcal and Computer Engneerng Georga Insttute of Technology, USA 2nd IEEE Internatonal
More informationDynamic Programming. Preview. Dynamic Programming. Dynamic Programming. Dynamic Programming (Example: Fibonacci Sequence)
/24/27 Prevew Fbonacc Sequence Longest Common Subsequence Dynamc programmng s a method for solvng complex problems by breakng them down nto smpler sub-problems. It s applcable to problems exhbtng the propertes
More informationWinter 2008 CS567 Stochastic Linear/Integer Programming Guest Lecturer: Xu, Huan
Wnter 2008 CS567 Stochastc Lnear/Integer Programmng Guest Lecturer: Xu, Huan Class 2: More Modelng Examples 1 Capacty Expanson Capacty expanson models optmal choces of the tmng and levels of nvestments
More information(1, T) policy for a Two-echelon Inventory System with Perishableon-the-Shelf
Journal of Optmzaton n Industral Engneerng 6 (24) 3-4 (, ) polcy for a wo-echelon Inventory ystem wth Pershableon-the-helf Items Anwar Mahmood a,, Alreza Ha b a PhD Canddate, Industral Engneerng, Department
More informationApplied Stochastic Processes
STAT455/855 Fall 23 Appled Stochastc Processes Fnal Exam, Bref Solutons 1. (15 marks) (a) (7 marks) The dstrbuton of Y s gven by ( ) ( ) y 2 1 5 P (Y y) for y 2, 3,... The above follows because each of
More informationMinimizing Maximum Flow-time on Related Machines
Mnmzng Maxmum Flow-tme on Related Machnes Nkhl Bansal and Bouke Cloostermans Endhoven Unversty of Technology P.O. Box 513, 5600 Endhoven, The Netherlands b.cloostermans,n.bansal@tue.nl Abstract We consder
More informationA NOTE ON A PERIODIC REVIEW INVENTORY MODEL WITH UNCERTAIN DEMAND IN A RANDOM ENVIRONMENT. Hirotaka Matsumoto and Yoshio Tabata
Scentae Mathematcae Japoncae Onlne, Vol. 9, (23), 419 429 419 A NOTE ON A PERIODIC REVIEW INVENTORY MODEL WITH UNCERTAIN DEMAND IN A RANDOM ENVIRONMENT Hrotaka Matsumoto and Yosho Tabata Receved September
More informationOn the Multicriteria Integer Network Flow Problem
BULGARIAN ACADEMY OF SCIENCES CYBERNETICS AND INFORMATION TECHNOLOGIES Volume 5, No 2 Sofa 2005 On the Multcrtera Integer Network Flow Problem Vassl Vasslev, Marana Nkolova, Maryana Vassleva Insttute of
More informationOrder Optimal Delay for Opportunistic Scheduling in Multi-User Wireless Uplinks and Downlinks
IEEE TRANSACTIONS ON NETWORKING, VOL. 16, NO. 5, PP. 1188-1199, OCT. 008 1 Order Optmal Delay for Opportunstc Schedulng n Mult-User Wreless Uplnks and Downlnks Mchael J. Neely Unversty of Southern Calforna
More informationLecture Notes on Linear Regression
Lecture Notes on Lnear Regresson Feng L fl@sdueducn Shandong Unversty, Chna Lnear Regresson Problem In regresson problem, we am at predct a contnuous target value gven an nput feature vector We assume
More informationRisk-Averse Group Elevator Scheduling
MITSUBISHI ELECTRIC RESEARCH LABORATORIES http://www.merl.com Rsk-Averse Group Elevator Schedulng Matthew Brand Danel Nkovsk TR004-45 May 004 Abstract We ntroduce a novel group elevator scheduler based
More informationGeneralized Linear Methods
Generalzed Lnear Methods 1 Introducton In the Ensemble Methods the general dea s that usng a combnaton of several weak learner one could make a better learner. More formally, assume that we have a set
More informationScheduling in polling systems
Schedulng n pollng systems Adam Werman, Erk M.M. Wnands, and Onno J. Boxma Abstract: We present a smple mean value analyss MVA framework for analyzng the effect of schedulng wthn queues n classcal asymmetrc
More informationThe Value of Demand Postponement under Demand Uncertainty
Recent Researches n Appled Mathematcs, Smulaton and Modellng The Value of emand Postponement under emand Uncertanty Rawee Suwandechocha Abstract Resource or capacty nvestment has a hgh mpact on the frm
More information1 Introduction 3. 2 Preliminaries A GPS server with variable service rate Leaky bucket... 4
Abstract Future ntegrated-servces packet networks wll carry a wde range of applcatons whch could dffer sgnfcantly n ther Qualty-of-Servce (QoS) requrements. Generalzed processor sharng (GPS) s the most
More information