Cognitive electronic warfare : an artificial intelligence approach

This comprehensive book gives an overview of how cognitive systems and artificial intelligence (AI) can be used in electronic warfare (EW). Readers will learn how EW systems respond more quickly and effectively to battlefield conditions where sophisticated radars and spectrum congestion put a high priority on EW systems that can characterize and classify novel waveforms, discern intent, and devise and test countermeasures. Specific techniques are covered for optimizing a cognitive EW system as well as evaluating its ability to learn new information in real time. The book presents AI for electronic support (ES), including characterization, classification, patterns of life, and intent recognition. Optimization techniques, including temporal tradeoffs and distributed optimization challenges are also discussed. The issues concerning real-time in-mission machine learning and suggests some approaches to address this important challenge are presented and described. The book covers electronic battle management, data management, and knowledge sharing. Evaluation approaches, including how to show that a machine learning system can learn how to handle novel environments, are also discussed. Written by experts with first-hand experience in AI-based EW, this is the first book on in-mission real-time learning and optimization. Cognitive Electronic Warfare: An Artificial Intelligence Approach Contents Foreword Preface 1 Introduction to Cognitive EW 1.1 What Makes a Cognitive System? 1.2 A Brief Introduction to EW 1.3 EW Domain Challenges Viewed from an AI Perspective 1.3.1 SA for ES and EW BDA 1.3.2 DM for EA, EP, and EBM 1.3.3 User Requirements 1.3.4 Connection between CR and EW Systems 1.3.5 EW System Design Questions 1.4 Choices: AI or Traditional? 1.5 Reader’s Guide 1.6 Conclusion References 2 Objective Function 2.1 Observables That Describe the Environment 2.1.1 Clustering Environments 2.2 Control Parameters to Change Behavior 2.3 Metrics to Evaluate Performance 2.4 Creating a Utility Function 2.5 Utility Function Design Considerations 2.6 Conclusion References 3 ML Primer 3.1 Common ML Algorithms 3.1.1 SVMs 3.1.2 ANNs 3.2 Ensemble Methods 3.3 Hybrid ML 3.4 Open-Set Classification 3.5 Generalization and Meta-learning 3.6 Algorithmic Trade-Offs 3.7 Conclusion References 4 Electronic Support 4.1 Emitter Classification and Characterization 4.1.1 Feature Engineering and Behavior Characterization 4.1.2 Waveform Classification 4.1.3 SEI 4.2 Performance Estimation 4.3 Multi-Intelligence Data Fusion 4.3.1 Data Fusion Approaches 4.3.2 Example: 5G Multi-INT Data Fusion for Localization 4.3.3 Distributed-Data Fusion 4.4 Anomaly Detection 4.5 Causal Relationships 4.6 Intent Recognition 4.6.1 Automatic Target Recognition and Tracking 4.7 Conclusion References 5 EP and EA 5.1 Optimization 5.1.1 Multi-Objective Optimization 5.1.2 Searching Through the Performance Landscape 5.1.3 Optimization Metalearning 5.2 Scheduling 5.3 Anytime Algorithms 5.4 Distributed Optimization 5.5 Conclusion References 6 EBM 6.1 Planning 6.1.1 Planning Basics: Problem Definition, and Search 6.1.2 Hierarchical Task Networks 6.1.3 Action Uncertainty 6.1.4 Information Uncertainty 6.1.5 Temporal Planning and Resource Management 6.1.6 Multiple Timescales 6.2 Game Theory 6.3 HMI 6.4 Conclusion References 7 Real-Time In-mission Planning and Learning 7.1 Execution Monitoring 7.1.1 EW BDA 7.2 In-Mission Replanning 7.3 In-Mission Learning 7.3.1 Cognitive Architectures 7.3.2 Neural Networks 7.3.3 SVMs 7.3.4 Multiarmed Bandi 7.3.5 MDPs 7.3.6 Deep Q-Learning 7.4 Conclusion References 8 Data Management 8.1 Data Management Process 8.1.1 Metadata 8.1.2 Semantics 8.1.3 Traceability 8.2 Curation and Bias 8.3 Data Management 8.3.1 Data in an Embedded System 8.3.2 Data Diversity 8.3.3 Data Augmentation 8.3.4 Forgetting 8.3.5 Data Security 8.4 Conclusion References 9 Architecture 9.1 Software Architecture: Interprocess 9.2 Software Architecture: Intraprocess 9.3 Hardware Choices 9.4 Conclusion References 10 Test and Evaluation 10.1 Scenario Driver 10.2 Ablation Testing 10.3 Computing Accuracy 10.3.1 Regression and Normalized RMSE 10.3.2 Classification and Confusion Matrices 10.3.3 Evaluating Strategy Performance 10.4 Learning Assurance: Evaluating a Cognitive System 10.4.1 Learning Assurance Process 10.4.2 Formal Verification Methods 10.4.3 Empirical and Semiformal Verification Methods 10.5 Conclusion References 11 Getting Started: First Steps 11.1 Development Considerations 11.2 Tools and Data 11.2.1 ML Toolkits 11.2.2 ML Datasets 11.2.3 RF Data-Generation Tools 11.3 Conclusion References Acronyms About the Authors Index