Skip to:Content
|
Bottom
Cover image for Feedback control of computing systems
Title:
Feedback control of computing systems
Publication Information:
Hoboken, NJ : John Wiley & Sons, 2004
ISBN:
9780471266372

Available:*

Library
Item Barcode
Call Number
Material Type
Item Category 1
Status
Searching...
30000010082180 TJ216 F43 2004 Open Access Book Book
Searching...

On Order

Summary

Summary

This is the first practical treatment of the design and application of feedback control of computing systems. MATLAB files for the solution of problems and case studies accompany the text throughout. The book discusses information technology examples, such as maximizing the efficiency of Lotus Notes.
This book results from the authors' research into the use of control theory to model and control computing systems. This has important implications to the way engineers and researchers approach different resource management problems. This guide is well suited for professionals and researchers in information technology and computer science.


Author Notes

JOSEPH L. HELLERSTEIN, YIXIN DIAO, and SUJAY PAREKH are researchers at the IBM Thomas J. Watson Research Center in Hawthorne, New York. They are also adjunct professors at Columbia University, where they are using this book to teach a class on feedback control to computer science students.

DAWN M. TILBURY is Associate Professor of Mechanical Engineering at the University of Michigan.


Table of Contents

Prefacep. xv
Part I Backgroundp. 1
1 Introduction and Overviewp. 3
1.1 The Nature of Feedback Controlp. 3
1.2 Control Objectivesp. 6
1.3 Properties of Feedback Control Systemsp. 7
1.4 Open-Loop versus Closed-Loop Controlp. 10
1.5 Summary of Applications of Control Theory to Computing Systemsp. 11
1.6 Computer Examples of Feedback Control Systemsp. 13
1.6.1 IBM Lotus Domino Serverp. 13
1.6.2 Queueing Systemsp. 15
1.6.3 Apache HTTP Serverp. 16
1.6.4 Random Early Detection of Router Overloadsp. 19
1.6.5 Load Balancingp. 20
1.6.6 Streaming Mediap. 21
1.6.7 Caching with Differentiated Servicep. 22
1.7 Challenges in Applying Control Theory to Computing Systemsp. 24
1.8 Summaryp. 26
1.9 Exercisesp. 27
Part II System Modelingp. 29
2 Model Constructionp. 31
2.1 Basics of Queueing Theoryp. 31
2.2 Modeling Dynamic Behaviorp. 35
2.2.1 Model Variablesp. 35
2.2.2 Signalsp. 35
2.2.3 Linear, Time-Invariant Difference Equationsp. 38
2.2.4 Nonlinearitiesp. 40
2.3 First-Principles Modelsp. 42
2.4 Black-Box Modelsp. 44
2.4.1 Model Scopep. 45
2.4.2 Experimental Designp. 47
2.4.3 Parameter Estimationp. 49
2.4.4 Model Evaluationp. 53
2.5 Summaryp. 56
2.6 Extended Examplesp. 56
2.6.1 IBM Lotus Domino Serverp. 56
2.6.2 Apache HTTP Serverp. 57
2.6.3 M/M/1/K Comparisonsp. 58
2.7 Parameter Estimation Using MATLABp. 59
2.8 Exercisesp. 62
3 Z-Transforms and Transfer Functionsp. 65
3.1 Z-Transform Basicsp. 65
3.1.1 Z-Transform Definitionp. 66
3.1.2 Z-Transforms of Common Signalsp. 68
3.1.3 Properties of Z-Transformsp. 71
3.1.4 Inverse Z-Transformsp. 74
3.1.5 Using Z-Transforms to Solve Difference Equationsp. 75
3.2 Characteristics Inferred from Z-Transformsp. 81
3.2.1 Review of Complex Variablesp. 81
3.2.2 Poles and Zeros of a Z-Transformp. 83
3.2.3 Steady-State Analysisp. 86
3.2.4 Time Domain versus Z-Domainp. 88
3.3 Transfer Functionsp. 89
3.3.1 Stabilityp. 92
3.3.2 Steady-State Gainp. 95
3.3.3 System Orderp. 96
3.3.4 Dominant Poles and Model Simplificationp. 96
3.3.5 Simulating Transfer Functionsp. 100
3.4 Summaryp. 102
3.5 Extended Examplesp. 103
3.5.1 M/M/1/K from System Identificationp. 103
3.5.2 IBM Lotus Domino Server: Sensor Delayp. 103
3.5.3 Apache HTTP Server: Combining Control Inputsp. 104
3.6 Z-Transforms and MATLABp. 105
3.7 Exercisesp. 107
4 System Modeling with Block Diagramsp. 111
4.1 Block Diagrams Basicsp. 111
4.2 Transforming Block Diagramsp. 115
4.2.1 Special Aggregations of Blocksp. 115
4.3 Transfer Functions for Control Analysisp. 116
4.4 Block Diagram Restructuringp. 119
4.5 Summaryp. 120
4.6 Extended Examplesp. 121
4.6.1 IBM Lotus Domino Serverp. 121
4.6.2 Apache HTTP Server with Control Loopsp. 123
4.6.3 Streamingp. 124
4.7 Composing Transfer Functions in MATLABp. 126
4.8 Exercisesp. 128
5 First-Order Systemsp. 129
5.1 First-Order Modelp. 129
5.2 System Responsep. 131
5.2.1 Steady-State and Transient Responsesp. 131
5.2.2 Input Signal Modelp. 133
5.2.3 Time-Domain Solutionp. 133
5.3 Initial Condition Responsep. 135
5.4 Impulse Responsep. 136
5.5 Step Responsep. 141
5.5.1 Numerical Examplep. 141
5.5.2 Time-Domain Solutionp. 141
5.5.3 Steady-State Responsep. 143
5.5.4 Transient Responsep. 144
5.6 Transient Response to Other Signalsp. 147
5.6.1 Ramp Responsep. 147
5.6.2 Frequency Responsep. 150
5.7 Effect of Stochasticsp. 152
5.8 Summaryp. 154
5.9 Extended Examplesp. 156
5.9.1 Estimating Operating Region of the Apache HTTP Serverp. 156
5.9.2 IBM Lotus Domino Server with a Disturbancep. 157
5.9.3 Feedback Control of the IBM Lotus Domino Serverp. 159
5.10 Analyzing Transient Response with MATLABp. 161
5.11 Exercisesp. 162
6 Higher-Order Systemsp. 165
6.1 Motivation and Definitionsp. 165
6.2 Real Polesp. 168
6.2.1 Initial Condition Responsep. 168
6.2.2 Impulse Responsep. 171
6.2.3 Step Responsep. 174
6.2.4 Other Signalsp. 176
6.2.5 Effect of Zerosp. 177
6.3 Complex Polesp. 179
6.3.1 Second-Order Systemp. 179
6.3.2 Impulse Responsep. 181
6.3.3 Step Responsep. 183
6.4 Summaryp. 185
6.5 Extended Examplesp. 186
6.5.1 Apache HTTP Server with a Filterp. 186
6.5.2 Apache HTTP Server with a Filter and Controllerp. 189
6.5.3 IBM Lotus Domino Server with a Filter and Controllerp. 191
6.5.4 M/M/1/K with a Filter and Controllerp. 192
6.6 Analyzing Transient Response with MATLABp. 196
6.7 Exercisesp. 197
7 State-Space Modelsp. 201
7.1 State Variablesp. 201
7.2 State-Space Modelsp. 204
7.3 Solving Difference Equations in State Spacep. 207
7.4 Converting Between Transfer Function Models and State-Space Modelsp. 211
7.5 Analysis of State-Space Modelsp. 216
7.5.1 Stability Analysis of State-Space Modelsp. 216
7.5.2 Steady-State Analysis of State-Space Modelsp. 218
7.5.3 Transient Analysis of State-Space Modelsp. 220
7.6 Special Considerations in State-Space Modelsp. 221
7.6.1 Equivalence of State Variablesp. 221
7.6.2 Controllabilityp. 222
7.6.3 Observabilityp. 225
7.7 Summaryp. 228
7.8 Extended Examplesp. 229
7.8.1 MIMO System Identification of the Apache HTTP Serverp. 229
7.8.2 State-Space Model of the IBM Lotus Domino Server with Sensor Delayp. 234
7.9 Constructing State-Space Models in MATLABp. 237
7.10 Exercisesp. 239
Part III Control Analysis and Designp. 243
8 Proportional Controlp. 245
8.1 Control Laws and Controller Operationp. 245
8.2 Desirable Properties of Controllersp. 252
8.3 Framework for Analyzing Proportional Controlp. 254
8.3.1 Closed-Loop Transfer Functionsp. 255
8.3.2 Stabilityp. 257
8.3.3 Accuracyp. 258
8.3.4 Settling Timep. 260
8.3.5 Maximum Overshootp. 260
8.4 P-Control: Robustness, Delays, and Filtersp. 261
8.4.1 First-Order Target Systemp. 261
8.4.2 Measurement Delayp. 266
8.4.3 Moving-Average Filterp. 268
8.5 Design of Proportional Controllersp. 271
8.6 Summaryp. 275
8.7 Extended Examplesp. 276
8.7.1 IBM Lotus Domino Server with a Moving-Average Filterp. 276
8.7.2 Apache with Precompensationp. 278
8.7.3 Apache with Disturbance Rejectionp. 282
8.7.4 Effect of Operating Region on M/M/1/K Controlp. 282
8.8 Designing P-Controllers in MATLABp. 286
8.9 Exercisesp. 289
9 PID Controllersp. 293
9.1 Integral Controlp. 293
9.1.1 Steady-State Error with Integral Controlp. 294
9.1.2 Transient Response with Integral Controlp. 296
9.2 Proportional-Integral Controlp. 301
9.2.1 Steady-State Error with PI Controlp. 303
9.2.2 PI Control Design by Pole Placementp. 303
9.2.3 PI Control Design Using Root Locusp. 307
9.2.4 PI Control Design Using Empirical Methodsp. 309
9.3 Proportional-Derivative Controlp. 315
9.4 PID Controlp. 320
9.5 Summaryp. 324
9.6 Extended Examplesp. 325
9.6.1 PI Control of the Apache HTTP Server Using Empirical Methodsp. 325
9.6.2 Designing a PI Controller for the Apache HTTP Server Using Pole Placement Designp. 327
9.6.3 IBM Lotus Domino Server with a Sensor Delayp. 328
9.6.4 Caching with Feedback Controlp. 330
9.7 Designing PI Controllers in MATLABp. 332
9.8 Exercisesp. 333
10 State-Space Feedback Controlp. 337
10.1 State-Space Analysisp. 337
10.2 State Feedback Control Systemsp. 339
10.2.1 Static State Feedbackp. 340
10.2.2 Precompensated Static State Feedbackp. 342
10.2.3 Dynamic State Feedbackp. 346
10.2.4 Comparison of Control Architecturesp. 351
10.3 Design Techniquesp. 353
10.3.1 Pole Placement Designp. 353
10.3.2 LQR Optimal Control Designp. 358
10.4 Summaryp. 362
10.5 Extended Examplesp. 364
10.5.1 MIMO Control of the Apache HTTP Serverp. 364
10.5.2 Effect of the LQR Design Parameters in a Dynamic State Feedback Systemp. 370
10.6 Designing State-Space Controllers Using MATLABp. 372
10.7 Exercisesp. 373
11 Advanced Topicsp. 375
11.1 Motivating Examplep. 376
11.2 Gain Schedulingp. 378
11.3 Self-Tuning Regulatorsp. 381
11.4 Minimum-Variance Controlp. 384
11.5 Fluid Flow Analysisp. 386
11.6 Fuzzy Controlp. 389
11.7 Summaryp. 393
11.8 Exercisesp. 395
Appendix A Mathematical Notationp. 397
Appendix B Acronymsp. 401
Appendix C Key Resultsp. 403
C.1 Modelingp. 403
C.1.1 Dominant Pole Approximationp. 403
C.1.2 Closed-Loop Transfer Functionsp. 403
C.2 Analysisp. 404
C.2.1 Stabilityp. 404
C.2.2 Settling Timep. 405
C.2.3 Maximum Overshootp. 405
C.2.4 Steady-State Gainp. 405
C.3 Controller Designp. 405
C.3.1 Control Lawsp. 405
C.3.2 Pole Placement Designp. 406
C.3.3 LQR Designp. 407
Appendix D Essentials of Linear Algebrap. 409
D.1 Matrix Inverse, Singularityp. 409
D.2 Matrix Minor, Determinant, and Adjointp. 409
D.3 Vector Spacesp. 410
D.4 Matrix Rankp. 411
D.5 Eigenvaluesp. 411
Go to:Top of Page