Performance Evaluation Methodologies and Tools: 15th EAI International Conference, VALUETOOLS 2022, Virtual Event, November 2022, Proceedings

Description

This book constitutes the refereed conference proceedings of the 15th International Conference on Performance Evaluation Methodologies and Tools, VALUETOOLS 2022, held in November 2022. Due to the safety concerns and travel restrictions caused by COVID-19, VALUETOOLS 2022 took place online in a live stream. The conference provides a world-leading and multidisciplinary venue for researchers and practitioners in diverse disciplines such as computer science, networks and telecommunications, operations research, optimization, control theory and manufacturing. The 18 full papers were carefully reviewed and selected from 47 submissions and are grouped in thematically as following: game theory; queueing models; applications; retrial queues; performance analysis and networking; distributed computing. Preface Organization Contents Game Theory Strategic Customer Behaviors in Observable Multi-server Batch Service Queueing Systems with Shared Fee and Server Maintenance Cost 1 Introduction 2 Modelling 3 Numerical Examples 3.1 Fee (p) 3.2 Batch Size (K) 3.3 Number of Servers (n) 4 Conclusion References Learning a Correlated Equilibrium with Perturbed Regret Minimization 1 Introduction 1.1 Related Work 1.2 Outline 2 Preliminaries 2.1 Notations 2.2 Model 2.3 Correlated Equilibria 2.4 Regret Minimization and Correlated Equilibria 3 Learning Rule 3.1 Device 3.2 Players' Strategies 3.3 Players' Moods Dynamic 3.4 Adaptive CPRM 4 Numerical Results 4.1 Matrix Games 4.2 Congestion Game 5 Conclusion A Appendix: Numerical Implementation of CPRM References Cooperative Game Theoretic Analysis of Shared Services 1 Introduction 1.1 Related Literature 2 Model Description 2.1 N-class Game 2.2 Importance of Worth Functions in N-class Game 3 2-class Game Illustration 3.1 The Core 3.2 Shapley Value 3.3 Fair Scheduling Policies 3.4 Nucleolus 4 Discussion A Proofs References Queueing Models System Content Analysis for a Two-Class Queue Where Service Times in a Busy Period Depend on the Presence of Class-2 1 Introduction 2 Queueing Model and System Equations 3 Queueing Analysis 4 Numerical Results 5 Conclusions References Asymptotic Analysis of Modified Erlang-B System with Sensing Time and Stochastic Loss of Secondary Users 1 Introduction 2 Model Description 3 First-Order Approximation 4 Main Results 5 Numerical Experiment 6 Concluding Remark References A Product-Form Solution for a Two-Class GeoGeo/D/1 Queue with Random Routing and Randomly Alternating Service 1 Introduction 2 Mathematical Model 3 Preliminaries 3.1 The Functional Equation 3.2 The Marginal PGFs 4 Complex-Analytic Analysis 4.1 Regions of Convergence 4.2 Analysis of the Kernel K(z1,z2) 4.3 Analytic Continuation of U(z,0) and U(0,z) 4.4 The Joint PGF 4.5 The Steady-State Joint Probabilities 5 Intuitive Stochastic Analysis 6 Direct Extensions of the Queueing Model 6.1 Multiple Arrivals 6.2 Multiple Servers 6.3 Multiple Queues References Applications Short Term Wind Turbine Power Output Prediction 1 Introduction 1.1 Contribution of the Paper 1.2 Paper Outline 2 Power Curve Modeling and Its Limitations 2.1 Power Curve Modeling Classes 2.2 Comparison of Models from Different Classes 3 Including More Physical Parameters 4 Predicting Power Output Using Time Series 4.1 Dynamic Modeling 4.2 Comparison of Model's Forecasting Capabilities 4.3 Forecasting Confidence 5 Concluding Remarks A Appendix: Data A.1 Description of the Raw Dataset A.2 Cleaning the data References A Domain Specific Language for the Design of Artificial Intelligence Applications for Process Engineering 1 Introduction 2 The Application Domain 2.1 Process Measurements and Control 3 Related Works 3.1 AI and Chemical Process Plants 3.2 DSL for Performance Evaluation 4 Design of the PE-AI-Perf DSL 4.1 Process Description 4.2 Definition of the Elements of the DSL 5 A Case Study 6 Conclusion and Future Work References Regret-Optimal Online Caching for Adversarial and Stochastic Arrivals 1 Introduction 1.1 Our Contributions 2 Problem Formulation 3 Policies 3.1 Least Frequently Used (LFU) 3.2 Follow the Perturbed Leader (FTPL) 3.3 Wait Then FTPL (W-FTPL) 4 Setting 1: Unrestricted Switching with Switching Cost 4.1 Adversarial Requests 4.2 Stochastic Requests 5 Restricted Switching 5.1 Stochastic Requests 5.2 Adversarial Requests 6 Numerical Experiments 6.1 Setting 1 with Stochastic File Requests 6.2 Setting 1 with Adversarial File Requests 6.3 Setting 2 7 Conclusion References Retrial Queues Cyclic Retrial Queue for Building Data Transmission Networks 1 Introduction 2 Mathematical Model 2.1 Modeling of Server Vacations 3 Kolmogorov Equations 4 First Step of Asymptotic-Diffusion Analysis 5 Second Step of Asymptotic-Diffusion Analysis 6 Algorithm of Calculation Probability Distribution Pdif(i) 7 Domain of Applicability of the Asymptotic-Diffusion Analysis 8 Conclusion References The Asymmetric Join the Shortest Orbit Queue: Analysis via a Dimensionality Reduction 1 Introduction 2 The System Model and the Equilibrium Equations 2.1 Preliminary Results 2.2 Dimensionality Reduction 3 The Compensation Method 4 Discussion on the Case of a Priority Line of a Finite Capacity 5 Conclusion and Future Work References Performance Analysis Understanding Slowdown in Large-Scale Heterogeneous Systems 1 Introduction 2 Model and Preliminaries 3 Analysis 4 Results and Discussion 4.1 Structure of Optimal PSITA Policies 4.2 Tradeoff Between E[S] and E[T] 5 Conclusion References Multi-Model Federated Learning with Provable Guarantees 1 Introduction 1.1 Our Contributions 1.2 Related Work 2 Setting 2.1 Performance Metric 3 Algorithms 3.1 Multi-FedAvg-Random (MFA-Rand) 3.2 Multi-FedAvg-Round Robin (MFA-RR) 4 Convergence of MFA-Rand and MFA-RR 5 Advantage of Multi-Model FL over Single Model FL 6 Simulations in Strongly Convex Setting 6.1 Simulation Framework 6.2 Comparison of MFA-Rand and MFA-RR 6.3 Gain Vs M 7 Gain Results on Synthetic and Real Federated Datasets 7.1 Gain Vs M 7.2 Real-world Scenarios 8 Conclusions References A Large Deviations Model for Latency Outage for URLLC 1 Introduction 1.1 Context 1.2 Related Works 1.3 Paper Organization 2 Outage Model 2.1 System and Traffic Model 2.2 Outage Bounds for a Tight Delay Budget (No Waiting) 2.3 Model with Waiting 3 Numerical Applications 3.1 Model with No Waiting 3.2 Model with Queuing Delay 4 Resource Dimensioning Framework 4.1 Architecture 4.2 System Level Simulation Results 5 Conclusion References Networking and Distributed Computing Some Basic Properties of Length Rate Quotient 1 Introduction 2 The LRQ Algorithm and Preliminaries 2.1 The LRQ Interleaved Shaping Algorithm 2.2 Flow and Server Models 3 Basic Properties of Interleaved LRQ Shapers 3.1 The ``Shaping-for-Free'' Property 3.2 Basic Properties of a Standalone LRQ Interleaved Shaper 4 Conclusion References Robustness of the Tangle 2.0 Consensus 1 Introduction 1.1 Results and Contribution 1.2 Structure of the Paper 2 Fundamentals of the Protocol 2.1 The UTXO Ledger and Conflicts 2.2 The Tangle 2.3 Sybil Protection 2.4 Approval Weight, Voting, and Confirmation 2.5 Liveness and Safety 2.6 Synchronised Random Reality Selection 3 The Adversary Model 3.1 Bait-and-Switch Attack 4 Simulation Components 4.1 Conflicts and Colours 4.2 Communication Layer 4.3 Agent-based Node Model 4.4 Weight Distribution 4.5 SRRS Implementation 5 Simulation Setup 6 Simulation Results 6.1 OTV Protocol Without SRRS 6.2 OTV Protocol with SRRS 6.3 Confirmation Time Analysis 7 Conclusion References Renting Edge Computing Resources for Service Hosting 1 Introduction 1.1 Our Contributions 1.2 Related Work 2 Setting 2.1 Cost Model and Constraints 2.2 Performance Metrics 3 Policies 3.1 Better Late Than Never (BLTN) 3.2 Follow the Perturbed Leader (FTPL) 3.3 An Illustration 4 Main Results and Discussion 5 Simulations 6 Conclusion References TAWNS – A Terrestrial Acoustic and Wireless Network Simulation Framework*-12pt 1 Introduction 2 Related Work 2.1 Wireless Network Simulators 2.2 Generating Spatial Audio Mixtures 3 Requirement Analysis 4 System Architecture 4.1 Sound Attenuation Model of Scaper-Extension 5 Simulation Performance 5.1 Run-Time 5.2 Memory 5.3 Disk Capacity 6 Conclusion References Author Index