| Preface | p. V |
| Ken Sevcik as an Advisor and Mentor | p. 1 |
| Shadow Servers and Priority Scheduling | p. 47 |
| Introduction | p. 7 |
| Single Class Models | p. 8 |
| Multi-Class Models | p. 8 |
| Importance of Priorities | p. 9 |
| The Shadow Server Approximation | p. 10 |
| Extensions | p. 12 |
| Comments on Significance | p. 13 |
| References | p. 14 |
| On the Chronology of Dynamic Allocation Index Policies: The Pioneering Work of K. C. Sevcik | p. 15 |
| Introduction | p. 15 |
| Sevcik's Smallest-Rank-First Index Policy | p. 16 |
| Background and Chronology | p. 17 |
| Examples | p. 18 |
| Concluding Remarks | p. 19 |
| References | p. 19 |
| Operational Analysis | p. 21 |
| Introduction | p. 21 |
| Dead Cows | p. 21 |
| Dead Cows in Markovian Queueing Networks | p. 22 |
| The Birth of Operational Analysis | p. 24 |
| The Fundamental Assumptions of Operational Analysis | p. 25 |
| Controversy | p. 28 |
| Salute | p. 29 |
| An Historical Footnote | p. 29 |
| References (Published) | p. 29 |
| References (Unpublished Technical Reports) | p. 30 |
| Operational Analysis: A Fable | p. 31 |
| Function Approximation by Random Neural Networks with a Bounded Number of Layers | p. 35 |
| Introduction | p. 35 |
| The GNN and Its Extensions | p. 36 |
| The BGNN model | p. 38 |
| Approximation of Functions of One Variable by the GNN with a Bounded Number of Layers | p. 40 |
| Technical premises | p. 41 |
| BGNN approximation of continuous functions of one variable | p. 44 |
| CGNN approximation of continuous functions of one variable | p. 46 |
| Approximation of Continuous Functions of s Variables | p. 49 |
| Conclusions | p. 53 |
| References | p. 54 |
| Proof of Technical Lemmas | p. 56 |
| The Achilles' Heel of Computer Performance Modeling and the Model Building Shield | p. 59 |
| Introduction | p. 59 |
| The Current Status of Model Building | p. 60 |
| System Multilevel Description | p. 61 |
| The system vertical description | p. 62 |
| The system horizontal description | p. 63 |
| The system software description | p. 64 |
| The Multilevel Model Building Method | p. 68 |
| The top-down bottom-up process | p. 68 |
| Comparison with Existing Approaches | p. 71 |
| Conclusions | p. 72 |
| Acknowledgment | p. 72 |
| References | p. 73 |
| Wireless Network Simulation: Towards a Systematic Approach | p. 75 |
| Introduction | p. 76 |
| Background and Model | p. 78 |
| Description of Our Framework | p. 79 |
| System parameters | p. 79 |
| Performance metrics | p. 80 |
| Our framework | p. 81 |
| Experimental Results | p. 82 |
| Parameters that affect steady state utilization | p. 82 |
| The significance of steady state arrival rate | p. 86 |
| Discussion and applications | p. 87 |
| Homogeneity | p. 90 |
| Related work | p. 91 |
| Metrics for comparison | p. 92 |
| Evaluation | p. 92 |
| Cell shape (number of neighbors) | p. 92 |
| User speed | p. 96 |
| User bandwidth requirement | p. 97 |
| Conclusion | p. 98 |
| References | p. 99 |
| Location- and Power-Aware Protocols for Wireless Networks with Asymmetric Links | p. 101 |
| Introduction and Motivation | p. 102 |
| Related Work | p. 104 |
| The Model of the System | p. 106 |
| m-Limited Forwarding | p. 110 |
| Simulation study | p. 112 |
| Routing Protocol | p. 118 |
| Neighbor discovery | p. 119 |
| Location and power update | p. 120 |
| Route discovery | p. 120 |
| Route maintenance | p. 121 |
| MAC Protocol | p. 121 |
| Topological considerations | p. 121 |
| A solution to the hidden node problem | p. 124 |
| Node status | p. 126 |
| Medium access model | p. 126 |
| A simulation study | p. 128 |
| Cross-Layer Architecture | p. 130 |
| Work in Progress | p. 131 |
| Summary | p. 132 |
| Acknowledgments | p. 133 |
| References | p. 133 |
| Multi-Threaded Servers with High Service Time Variation for Layered Queueing Networks | p. 137 |
| Introduction | p. 137 |
| Residence Time Expressions | p. 138 |
| MVA waiting time expressions | p. 139 |
| Accuracy and Computation-Time Comparisons | p. 141 |
| Example Case Studies | p. 143 |
| Systems management example | p. 143 |
| Electronic bookstore example | p. 144 |
| Conclusions | p. 148 |
| Acknowledgments | p. 150 |
| Marginal Probabilities | p. 150 |
| de Souza e Silva and Muntz Approximation | p. 152 |
| References | p. 152 |
| Quantiles of Sojourn Times | p. 155 |
| Introduction | p. 156 |
| Time Delays in the Single Server Queue | p. 158 |
| Waiting time distribution in the M/G/1 queue | p. 158 |
| Busy periods | p. 159 |
| Waiting times in LCFS queues | p. 160 |
| Waiting times with Processor-Sharing discipline | p. 162 |
| MM CPP/GE/c G-Queues: Semi-Numerical Laplace Transform Inversion | p. 162 |
| Time Delays in Networks of Queues | p. 166 |
| Open networks | p. 167 |
| Closed networks | p. 169 |
| Cyclic networks | p. 173 |
| Paths with service rates all equal | p. 174 |
| Passage Times in Continuous Time Markov Chains | p. 174 |
| First passage times in CTMCs | p. 174 |
| Uniformization | p. 175 |
| Hypergraph partitioning | p. 176 |
| Parallel algorithm and tool implementation | p. 177 |
| Numerical example | p. 179 |
| Passage Times in Continuous Time Semi-Markov Processes | p. 182 |
| First passage times in SMPs | p. 183 |
| Iterative passage time algorithm | p. 185 |
| Laplace transform inversion | p. 186 |
| Implementation | p. 187 |
| Numerical example | p. 187 |
| Conclusion | p. 190 |
| References | p. 191 |
| Asymptotic Solutions for Two Non-Stationary Problems in Internet Reliability | p. 195 |
| Introduction | p. 195 |
| Poisson Approximation for the Number of Failed Routers | p. 197 |
| Asymptotics of Lost Bandwidth | p. 200 |
| References | p. 204 |
| Burst Loss Probabilities in an OBS Network with Dynamic Simultaneous Link Possession | p. 205 |
| Introduction | p. 205 |
| Problem Description | p. 208 |
| A Queueing Network Model for an OBS Path | p. 209 |
| The arrival process | p. 211 |
| The Decomposition Algorithm | p. 213 |
| An example | p. 213 |
| Analysis of sub-system 1 | p. 213 |
| Analysis of sub-system 2 | p. 215 |
| The iterative procedure | p. 216 |
| The decomposition algorithm | p. 217 |
| Calculation of the burst loss probability | p. 219 |
| Numerical Results | p. 220 |
| Conclusions | p. 223 |
| References | p. 224 |
| Stochastic Analysis of Resource Allocation in Parallel Processing Systems | p. 227 |
| Introduction | p. 227 |
| Model of Parallel Processing Systems | p. 230 |
| Analysis of Dynamic Spacesharing | p. 232 |
| Irreducibility and stability criterion | p. 235 |
| Special case: Exponential model parameters | p. 235 |
| Performance measures | p. 237 |
| Analysis of Memory Reference Behavior | p. 239 |
| Program behavior models | p. 240 |
| Intra-locality memory overhead | p. 244 |
| Inter-locality memory overhead | p. 245 |
| Calculation of N[subscript I] | p. 246 |
| Calculation of C[subscript I] | p. 247 |
| Total memory overhead | p. 250 |
| Conclusions | p. 250 |
| Acknowledgment | p. 251 |
| References | p. 251 |
| Periodic Task Cluster Scheduling in Distributed Systems | p. 257 |
| Introduction | p. 257 |
| Model and Methodology | p. 260 |
| System and workload models | p. 260 |
| Scheduling strategies | p. 262 |
| Performance metrics | p. 263 |
| Model implementation and input parameters | p. 263 |
| Simulation Results and Performance Analysis | p. 264 |
| Normal distribution case | p. 264 |
| Uniform distribution case | p. 271 |
| Conclusions and Future Research | p. 273 |
| References | p. 273 |
| Table of Contents provided by Ingram. All Rights Reserved. |