[Lecture Notes in Computer Science] Measurement, Modelling and Evaluation of Computing Systems Volume 12040 (20th International GI/ITG Conference, MMB 2020, SaarbrÃ1⁄4cken, Germany, March 16–18, 2020, Proceedings) ||

This book constitutes the proceedings of the 20th International GI/ITG Conference on Measurement, Modelling and Evaluation of Computing Systems, MMB 2020, held in Saarbrücken, Germany, in March 2020. The 16 full papers presented in this volume were carefully reviewed and selected from 32 submissions. They are dealing with scientific aspects of measurement, modelling and evaluation of intelligent systems including computer architectures, communication networks, distributed systems and software, autonomous systems, workflow systems, cyber-physical systems and networks, Internet-of-Things, as well as highly dependable, highly performant and highly secure systems. Preface Organization Abstracts of Invited Talks Interference Networks Safety Certification of Deep Learning Predictable Latency in Softwarized Networks Contents Performance Analytics of a Virtual Reality Streaming Model 1 Introduction 2 Machine Learning and Performance Evaluation 2.1 Decision Trees 2.2 Model Trees 3 Stochastic Fluid Flow Model 3.1 Model Parameters 3.2 Fluid Flow Analysis of the Four-State System 3.3 Fluid Flow Analysis of the Two-State System 4 Steps in Performance Analytics 4.1 Data Generation 4.2 Data Preparation and Transformations 4.3 Implementation 4.4 Validation 5 Univariate Performance Analytics 5.1 Analytics of an RP/IRP/K System 5.2 Asymptotic Behaviours of RP/IRP/K Subsystems 6 Multivariate Performance Analytics 6.1 ARQ 6.2 Fading 7 Conclusions References To Fail or Not to Fail: Predicting Hard Disk Drive Failure Time Windows 1 Introduction 2 Background 2.1 Hard Disk Drive Monitoring 2.2 Random Forest Models 3 Related Work 4 Approaches for HDD Failure Level Prediction 4.1 Binary Classification of Failing HDDs 4.2 Classification of Multiple Failure Levels 4.3 Regression for Time-to-Failure Prediction 5 Evaluation 5.1 Evaluation Design 5.2 Binary Failure Prediction 5.3 Failure Level Classification 5.4 Time-to-Failure Regression 5.5 Training Time Comparison 6 Conclusion and Discussion References Concurrent MDPs with Finite Markovian Policies 1 Introduction 2 Background 2.1 Notation 2.2 Concurrent MDPs 2.3 Some Results for Concurrent MDPs 2.4 An Example 3 Computation of Optimal Policies 3.1 Integer Linear Programming 3.2 Policy Iteration 3.3 Value Iteration 4 Finite Expansion of Policies 4.1 Expansion of CMDPs 4.2 Matrix Structures 4.3 Policy Evaluation 4.4 Policy Iteration for Policies from PfMX 5 Examples 5.1 Random MDPs 5.2 Admission Control Problem 6 Conclusion References A Stochastic Automata Network Description for Spatial DNA-Methylation Models 1 Introduction 2 SANS 3 Results 4 Conclusion References An ns-3 Model for Multipath Communication with Terrestrialpg and Satellite Links 1 Introduction 1.1 Satellite Internet and Performance Enhancement Proxies 1.2 Multipath Communication 2 Related Work 3 The TMC Solution 4 Modeling 4.1 Traffic Generation and Workload Model 4.2 TMC Model 5 Experiment Setup 6 Evaluation 7 Conclusion and Future Work References Model-Based Performance Predictions for SDN-Based Networks: A Case Study 1 Introduction 2 Performance Modeling and Prediction Methodology 3 Case Study Design 3.1 Hardware Testbed 3.2 Modeling 4 Case Study 4.1 Non-SDN Networking 4.2 SDN Hardware Tables 4.3 Node Virtualization 4.4 Software Flow Tables 4.5 SDN Controller 5 Related Work 6 Conclusion References Design of a Hybrid Genetic Algorithm for Time-Sensitive Networking 1 Introduction 2 Related Work 3 Time-Sensitive Networking 3.1 Time-Aware Shaper 3.2 Scheduling Constraints 4 Genetic Algorithms 4.1 Chromosome Encoding for TSN 4.2 Population Initialization 4.3 Selection 4.4 Crossover Operators 4.5 Mutation 4.6 Replacement Strategies 4.7 GA Combined with Neighborhood Search 5 Application of the Hybrid Genetic Algorithm in TSN 5.1 GA Settings 5.2 Schedule Calculation 6 Computational Results 7 Conclusion References Performance Analysis for Loss Systems with Many Subscribers and Concurrent Services 1 Introduction 2 Setting of the General Problem 3 The Rough Model: Loss Probabilities 4 The Detailed Model: Revenue 5 Finite Population: Limited Number of Subscribers 6 Concluding Remarks References On the Stochastic End-to-End Delay Analysis in Sink Trees Under Independent and Dependent Arrivals 1 Introduction 1.1 Background 1.2 E2E Analysis in SNC – State of Affairs 1.3 Motivation and Contribution 1.4 Outline 2 SNC Background and Notation 3 Sink Tree End-to-End Delay Bound 3.1 Two-Server Sink Tree 3.2 The General Case 3.3 Delay Bounds with PMOO Under Dependent Cross-flows 4 Numerical Evaluation 4.1 Impact of a Finite Time Horizon 4.2 Comparison Between SFA and PMOO 4.3 Parameter Sensitivity of Fractional Brownian Motion 4.4 Scaling Effects of PMOO 4.5 Comparison Between Independent and Dependent Cross-flows 5 Conclusion A Appendix A.1 Proof of Proposition 3 References Graph-Based Mobility Models: Asymptotic and Stationary Node Distribution 1 Introduction 2 Pathway Mobility Models 2.1 The Graph for Graph-Based Mobility Models 3 Fixed Population: Closed Graph-Based Mobility Model 3.1 State Space for the Markov Model: Closed Area 3.2 Stationary Behaviour of Closed Graph-Based Mobility Models 4 Varying Population: Open Graph-Based Mobility Model 4.1 State Space for the Markov Model: Open Area 4.2 Stationary Behaviour of Open Graph-Based Mobility Models 5 Concluding Remarks References Parallelization of EM-Algorithms for Markovian Arrival Processes 1 Introduction 2 Background 2.1 Basic Definitions and Notation 2.2 Analysis of MAPs 3 EM-Algorithms for MAPs 3.1 Likelihood Optimization 3.2 An EM Algorithm 3.3 Implementation Issues 4 Parallel EM-Algorithm 4.1 Computation of Forward and Backward Vectors 4.2 Computation of the Flow Matrices 4.3 The Complete Algorithm 5 Examples 5.1 Results for the Trace BC-pAug89 5.2 Results for the Trace LBL-TCP3 5.3 Comparison with the Parallel EM Algorithm from ch11BHT18 6 Conclusion References It Sometimes Works: A Lifting Algorithm for Repair of Stochastic Process Algebra Models 1 Introduction 2 Modelling Framework and State of the Art 2.1 Stochastic Process Algebra (SPA) 2.2 Model Repair by Rate Modification 3 A Lifting Algorithm for Modular Model Repair 3.1 Motivating Examples 3.2 Idea of the Algorithm 3.3 Lifting Algorithm 3.4 Remarks on the Algorithm 3.5 Illustration of the Algorithm 4 Application to SPA Models with Product Form 5 Summary and Future Work References An Efficient Brute Force Approach to Fit Finite Mixture Distributions 1 Introduction 2 A General Brute Force Approach 2.1 Farey Sequences as a Basic Value Set 2.2 Transformation of Basic Value Set 2.3 Non-Negative Least Squares Problem Definition 3 Fitting Finite Mixtures of Erlang Distributions 4 Experimental Results 5 Extending the Approach 6 Conclusions References Freight Train Scheduling in Railway Systems 1 Introduction 2 Modeling 3 The Time-Dependent Shortest Path Problem 3.1 Dijkstra's Algorithm 3.2 Edge Restrictions 3.3 Vertex Restrictions 4 Experimental Results 5 Conclusion References A Tool for Requirements Analysis of Safety-Critical Cyber-Physical Systems 1 Introduction 2 A Generic Algorithm for Requirement Analysis 3 A Formal Model for Requirement Analysis of CPSs 4 An Implemented Algorithm for RA of SC-CPS 4.1 Capturing the Premature Requirements 4.2 Evolution of the Requirements 4.3 Validation of the Requirements 5 Experimental Results 6 Conclusion References Automated Rare Event Simulation for Fault Tree Analysis via Minimal Cut Sets 1 Introduction 2 Theoretical Framework 2.1 Fault Trees 2.2 Fault Tree Analysis via (Rare Event) Simulation 3 RES Using Minimal Cut Sets 3.1 MCS Re-writing of Trees 3.2 Importance Functions from Minimal Cut Sets 4 Empirical Evaluation 4.1 Case Study and Experimental Setting 4.2 Results and Discussion 5 Conclusions A Galileo of the [3] dft B Importance Functions Used for HECS Experiments References Author Index This book constitutes the proceedings of the 20th International GI/ITG Conference on Measurement, Modelling and Evaluation of Computing Systems, MMB 2020, held in Saarbr赣ken, Germany, in March 2020. The 16 full papers presented in this volume were carefully reviewed and selected from 32 submissions. They are dealing with scientific aspects of measurement, modelling and evaluation of intelligent systems including computer architectures, communication networks, distributed systems and software, autonomous systems, workflow systems, cyber-physical systems and networks, Internet-of-Things, as well as highly dependable, highly performant and highly secure systems