Computer Safety, Reliability, and Security : 40th International Conference, SAFECOMP 2021, York, UK, September 8–10, 2021, Proceedings

This book constitutes the proceedings of the 40th International Conference on Computer Safety, Reliability and Security, SAFECOMP 2021, which took place in York, UK, in September 2021. The 17 full papers included in this volume were carefully reviewed and selected from 76 submissions. They were organized in topical sections as follows: machine learning safety assurance; security engineering; safety and assurance cases; machine learning applications; safety validation and simulation; and fault tolerance. Preface Organization Contents Machine Learning Safety Assurance Evaluation Framework for Performance Limitation of Autonomous Systems Under Sensor Attack 1 Introduction 2 Evaluation Framework Based on SOTIF Process 2.1 Relevant Standards and SOTIF 2.2 Evaluation Framework 3 Identifying Attack Scenarios Using STAMP/STPA 3.1 STAMP/STPA Safety Analysis 3.2 Analysis Steps and Results 4 Evaluating Performance Limitations Under Sensor Attacks 4.1 Test Model for Verification of Safety Constraints 4.2 Sensor Attack Simulator 4.3 Evaluation Examples Using the Prototype 5 Related Work 6 Conclusion References Could We Relieve AI/ML Models of the Responsibility of Providing Dependable Uncertainty Estimates? A Study on Outside-Model Uncertainty Estimates 1 Introduction 2 Related Work on Uncertainty Predictions 3 Study Planning and Execution 3.1 Research Questions 3.2 Study Design and Variation Points 3.3 Study Execution 4 Study Results and Discussion 4.1 RQ1: Comparing UW Performance with In-Model Approaches 4.2 RQ2: Synergies Between In-Model and Outside-model Approaches 4.3 RQ3: Performance Under Common, Less Than Optimal Conditions 5 Conclusion References Towards Certification of a Reduced Footprint ACAS-Xu System: A Hybrid ML-Based Solution 1 Introduction 1.1 ACAS Xu Overview 1.2 Purpose of the Work 2 ACAS Xu Hybrid Architecture 2.1 Learning Process 2.2 Design of the Hybrid Architecture 2.3 Why a New Hybrid Architecture 3 Certification Methodology 3.1 Notations 3.2 Assurance Case for the Hybrid Controller 4 Related Work 5 Conclusion References Security Engineering IT Design for Resiliency Using Extreme Value Analysis 1 Introduction 2 Extreme Value Analysis 2.1 Detecting Extremity 3 Modeling Risk 3.1 Probabilistic Modeling for a Mission Duration 3.2 Mission Risk: The Cumulative Cost of Failures 3.3 Workflow 4 Case Study 4.1 Measurement-Based Extreme Value Analysis 4.2 Risk Model and Availability Evaluation 5 Extension to Cold-Backup 6 Conclusion References DeepCert: Verification of Contextually Relevant Robustness for Neural Network Image Classifiers 1 Introduction 2 DeepCert Verification Method 2.1 Overview 2.2 DeepCert Instantiation for Test-Based Verification 2.3 DeepCert Instantiation for Formal Verification 3 Implementation 4 Experimental Results 4.1 Case Study 1: Road Traffic Speed Sign Classification 4.2 Case Study 2: CIFAR-10 5 Related Work 6 Conclusions and Future Work References ISO/SAE 21434-Based Risk Assessment of Security Incidents in Automated Road Vehicles 1 Introduction 2 Related Work 3 Introduction to ISO/SAE 21434 4 Context-Aware Risk Assessments of Security Incidents 4.1 Offline Phase 4.2 Online Phase 5 Case Study and Discussion 5.1 Reference Vehicle 5.2 Application of the Offline Phase 5.3 Discussion of the Online Phase 6 Conclusion References Safety and Assurance Cases Automating the Assembly of Security Assurance Case Fragments 1 Introduction 2 Contribution 2.1 Augmented Goal Structuring Notation (GSN) 2.2 Security Assurance Case Patterns 2.3 Model-Based Architecture Analysis in VERDICT 2.4 Security Assurance Case Construction 3 Case Study: A Delivery Drone 4 Discussion 5 Related Work 6 Conclusion and Future Work References Safety Case Maintenance: A Systematic Literature Review 1 Introduction 2 Background and Related Work 3 Review Protocol 3.1 Establishing the Quasi-Gold Standards by Manual Search 3.2 Automated Search and Snowballing 3.3 Exclusion and Inclusion Criteria 3.4 Evaluation Criteria 4 Review Results 5 Discussion 6 Summary and Future Lines of Work References Towards Certified Analysis of Software Product Line Safety Cases 1 Introduction 2 Background 2.1 Safety Cases, GSN, and Change Impact Assessment 2.2 Software Product Lines 3 Methodology and Infrastructure 4 Change Impact Assessment 4.1 Single-Product Algorithm 4.2 Lifted Algorithm 4.3 Examples 5 Towards Implementation 6 Related Work 7 Conclusion and Future Work References Machine Learning Applications Safety Assurance of Machine Learning for Chassis Control Functions 1 Introduction 2 Related Work 3 Case Study 4 Assurance Approach 4.1 Domain Analysis 4.2 System Design 4.3 Verification and Validation 4.4 Assurance Case 5 Detailed Analysis of the Machine Learning Function 6 Lessons Learned 7 Conclusion References Safe Interaction of Automated Forklifts and Humans at Blind Corners in a Warehouse with Infrastructure Sensors 1 Introduction 2 Blind Corners in Warehouses 2.1 Definition of Blind Corners 2.2 Safety Standards for Driverless Industrial Trucks 2.3 Intersection Cooperation and Coordination 3 Infrastructure-Cooperative Autonomous Control 3.1 Infrastructure-Cooperative Autonomous Control Architecture 3.2 Infrastructure-Cooperative Autonomous Control Hazards 3.3 Safety Concepts for Safe Interaction of Automated Forklifts and Human Workers at Blind Corners 4 Evaluation 5 Conclusion and Outlook References Machine Learning-Based Fault Injection for Hazard Analysis and Risk Assessment 1 Introduction 2 Adaptive Cruise Control 3 ML-Based FI Within the SAHARA Methodology 3.1 Required Information 3.2 Scenario Selection 3.3 Fault Injection and Reinforcement Learning Algorithm 3.4 Simulation and Visualisation 3.5 Hazard Classification 4 Results 5 Discussion 6 Related Work 7 Conclusion and Future Work References Safety Validation and Simulation SASSI: Safety Analysis Using Simulation-Based Situation Coverage for Cobot Systems 1 Introduction 2 Case Study: Industrial Manufacturing Cobot 3 Overview of the SASSI Method 4 Analysing the System Safety 5 Monitoring Safety Artefacts 6 Generating and Evaluating Configurations 7 Evaluation 7.1 Problem Space 7.2 Simulation Setup 7.3 Search Heuristics 7.4 Results 8 Related Work 9 Conclusion References Attack and Fault Injection in Self-driving Agents on the Carla Simulator – Experience Report 1 Introduction 2 Background 2.1 Adversarial Attacks Using the ART Toolbox 2.2 Fault Injection in Trained Agents with PytorchFI 2.3 Carla Simulator and Learning by Cheating (LbC) 3 Injection of Attacks in a Self-driving Agent 3.1 Selection of Suitable Attacks 3.2 Integration of ART in Learning by Cheating 4 Injection of Faults in a Self-driving Agent 4.1 Perturbation Models 4.2 Application of PyTorchFi in LbC 5 Experiments and Results 5.1 Description of the Experimental Campaign 5.2 Adversarial Attacks Injection: Results 5.3 Faults Injection: Results 6 Limitations and Considerations for Real-World Scenarios 7 Conclusions References A Framework for Automated Quality Assurance and Documentation for Pharma 4.0 1 Introduction 2 State-of-the-Art of Quality Assurance and Documentation in the Pharmaceutical Industry 2.1 Lifecycle of Pharmaceutical Products and Good-X-Practice 2.2 The Patient Batch Record 2.3 Pre-Market Authorization, Research and Development 2.4 Production of Products with Market Authorization 3 Tools for Model-Based Quality Assurance and Dependability 3.1 Assurance Cases and the Digital Dependability Identity 3.2 Asset Administration Shells and Submodels 4 Framework for End-to-End Automated Quality Assurance and Documentation 4.1 A Car-T Production Process 4.2 A Car-T Cell Assurance Case 4.3 An Auto-Generated Batch Record Submodel 4.4 Submodels for Collaborating Assets 5 Conclusion References Fault Tolerance A Modular Approach to Non-deterministic Dynamic Fault Trees 1 Introduction 2 Related Work 3 Background 3.1 Fault Trees 3.2 Non-deterministic Dynamic Fault Trees 3.3 Synthesizing Recovery Strategies 4 Modular Synthesis of Recovery Automata 4.1 Modular Workflow 4.2 Modularization 5 Case Studies 6 Conclusions and Future Work References Composition of Fault Forests 1 Introduction 2 Running Example 3 Formalization 4 Implementation 4.1 Formal Background 4.2 Algorithm Implementation 5 Related Work 6 Conclusion and Future Work References Author Index This book constitutes the refereed proceedings of the 32nd International Conference on Computer Safety, Reliability, and Security, SAFECOMP 2013, held in Toulouse, France, in September 2013. The 20 revised full papers presented together with 5 practical experience reports were carefully reviewed and selected from more than 88 submissions. The papers are organized in topical sections on safety requirements and assurance, testing and verification, security, software reliability assessment, practical experience reports and tools, safety assurance in automotive, error control codes, dependable user interfaces, and hazard and failure mode analysis.