Network Security A Decision and Game-Theoretic Approach

Transcription

1 Network Security A Decision and Game-Theoretic Approach Tansu Alpcan Deutsche Telekom Laboratories, Technical University of Berlin, Germany and Tamer Ba ar University of Illinois at Urbana-Champaign, USA Щ CAMBRIDGE UNIVERSITY PRESS

2 Contents Preface Acknowledgments Artwork Notation page xii xv xvi xvii Part I Introduction Introduction Network security The approach Motivating examples Security games Security risk-management Optimal malware epidemic response Discussion and further reading 18 Network security concepts Networks and security threats Networks and the World Wide Web Security threats Attackers, defenders, and their motives Attackers Defenders Defense mechanisms Security tradeoffs and risk-management Security tradeoffs Security risk-management Discussion and further reading 34

3 viii Contents Part II Security games 37 3 Deterministic security games Security game model Intrusion detection games Matrix games Games with dynamic information Sensitivity analysis Modeling malicious behavior in social networks Security games for vehicular networks Vehicular network model Attack and defense model Game formulation and numerical analysis Security games in wireless networks Random access security games Interference limited multiple access security games Revocation games Revocation game model Sequential revocation games Static revocation games Discussion and further reading 72 4 Stochastic security games Markov security games Markov game model Solving Markov games Stochastic intrusion detection game Security of interconnected systems Analysis of an illustrative example Linear influence models Malware filter placement game Stochastic game formulation Simulations Discussion and further reading 95 5 Security games with information limitations Bayesian security games Bayesian intrusion detection game Bayesian games for wireless security Security games with observation and decision errors Game model and fictitious play Fictitious play with observation errors Fictitious play with decision errors 120

4 5.2.4 Time-invariant and adaptive fictitious play 5.3 Discussion and further reading Part III Decision making for network security 6 Security risk-management 6.1 Quantitative risk-management Risk in networked systems and organizations A probabilistic risk framework Dynamic risk mitigation and control 6.2 Security investment games Influence network and game model Equilibrium and convergence analysis Incentives and game design 6.3 Cooperative games for security risk-management Coalitional game model Coalition formation under ideal cooperation 6.4 Discussion and further reading 7 Resource allocation for security 7.1 An optimization approach to malware filtering Traffic centrality measures Filtering problem formulations 7.2 A robust control framework for security response Network traffic filtering model Derivation of optimal controller and state estimator 7.3 Optimal and robust epidemic response Epidemic models Feedback response for malware removal Multiple networks 7.4 Discussion and further reading 8 Usability, trust, and privacy 8.1 Security and usability A system for security alert dissemination Effective administrator response 8.2 Digital trust in online communities Community trust game Dynamics and convergence Numerical analysis 8.3 Location privacy in mobile networks A locati on privacy model

5 x Contents Location privacy games Discussion and further reading 215 Part IV Security attack and intrusion detection Machine learning for intrusion and anomaly detection Intrusion and anomaly detection Intrusion detection and prevention systems Open problems and challenges Machine learning for security: an overview Overview of machine-learning methods Open problems and challenges Distributed machine learning SVM classification and decomposition Parallel update algorithms Active set method and a numerical example Behavioral malware detection for mobile devices Discussion and further reading Hypothesis testing for attack detection Hypothesis testing and network security An overview of hypothesis testing Вayesian hypothesis testing Minimax hypothesis testing Neyman-Pearson hypothesis testing Other hypothesis testing schemes Decentralized hypothesis testing with correlated observations Decentralized hypothesis testing Decision rules at the sensors and at the fusion center Decentralized Bayesian hypothesis testing Decentralized Neyman-Pearson hypothesis testing The majority vote versus the likelihood ratio test An algorithm to compute the optimal thresholds Discussion and further reading 273 A Optimization, game theory, and optimal and robust control 274 A.l Introduction to optimization 274 A. 1.1 Sets, spaces, and norms 274 A. 1.2 Functionals, continuity, and convexity 275 A. 1.3 Optimization of functionals 276 A.2 Introduction to noncooperative game theory 281 A.2.1 General formulation for noncooperative games and equilibrium solutions 282

6 Contents xi A.2.2 Existence of Nash and saddle-point equilibria in finite games 284 A.2.3 Existence and uniqueness of Nash and saddle-point equilibria in continuous-kernel (infinite) games 285 A.2.4 Online computation of Nash equilibrium policies 287 A.3 Introduction to optimal and robust control theory 288 A.3.1 Dynamic programming for discrete-time systems 289 A.3.2 Dynamic programming for continuous-time systems 291 A.3.3 The minimum principle 294 A.3.4 H -optimal control 296 References Index