复杂网络控制技术
2009-4
高等教育出版社
迈恩
562
无
电力网络、柔性制造、移动通信等等互连都形成了复杂网络。本书着眼于复杂网络系统中的共性问题，综合了作者多年来在该方向深入、系统的研究成果，给出了建立网络模型所需要的工具和哲学思想，详细具体地把握了其动力学本质，同时简明地揭示了有效控制的解决方案及其分析。本书内容分为三个部分：第一部分为建模与控制，第二部分为负荷调度，第三部分为稳定性及性能分析。 本书内容循序渐进，每章均附有习题并提供习题解答。基础章节部分要求读者具有随机过程和线性代数知识，适用于信息类、电子类等专业高年级本科生，高级部分则适用于研究生、研究人员和从业者。
Sean Meyn，伊利诺斯大学电子与计算机工程系教授，IEEE Fellow。担任系统与控制、应用概率等领域多个期刊的编委。与他人合著的图书Markov Chains and Stochastic Stability获1994年ORSA／TIMS最佳著作奖。在MIT4 UTRC等世界各地多个大学担任客座教授。他的研究兴趣包括随机过程、最优化、复杂网络以及信息论等。
List of IllustrationsPrefaceDedication1 Introduction 1.1 Networks in practice 1.2 Mathematical models 1.3 What do you need to know to read this book? 1.4 NotesPart I: Modeling and Control2 Examples 2.1 Modeling the single server queue 2.2 Klimov model 2.3 Capacity and queueing in communication systems 2.4 Multiple-access communication 2.5 Processor sharing model 2.6 Inventory model 2.7 Power transmission network 2.8 Optimization in a simple re-entrant line 2.9 Contention for resources and instability 2.10 Routing model 2.11 Braess' paradox 2.12 Notes3 The Single Server Queue 3.1 Representations 3.2 Approximations 3.3 Stability 3.4 Invariance equations 3.5 Big queues 3.6 Model selection 3.7 Notes Exercises4 Scheduling 4.1 Controlled random-walk model 4.2 Fluid model 4.3 Control techniques for the fluid model 4.4 Comparing fluid and stochastic models 4.5 Structure of optimal policies 4.6 Safety-stocks 4.7 Discrete review 4.8 MaxWeight and MinDrift 4.9 Perturbed value function 4.10 Notes Exercises~Part II: Workload5 Workload and Scheduling 5.1 Single server queue 5.2 Workload for the CRW scheduling model 5.3 Relaxations for the fluid model 5.4 Stochastic workload models 5.5 Pathwise optimality and workload 5.6 Hedging in networks 5.7 Notes Exercises6 Routing and Resource Pooling 6.1 Workload in general models 6.2 Resource pooling 6.3 Routing and workload 6.4 MaxWeight for routing and scheduling 6.5 Simultaneous resource possession 6.6 Workload relaxations 6.7 Relaxations and policy synthesis for stochastic models 6.8 Notes Exercises7 Demand 7.1 Network models 7.2 Transients 7.3 Workload relaxations 7.4 Hedging in a simple inventory model 7.5 Hedging in networks 7.6 Summary of steady-state control techniques 7.7 Notes ExercisesPart III: Stability and Performance8 Foster-Lyapunov Techniques 8.1 Lyapunov functions 8.2 Lyapunov functions for networks 8.3 Discrete review 8.4 MaxWeight 8.5 MaxWeight and the average-cost optimality equation 8.6 Linear programs for performance bounds 8.7 Brownian workload model 8.8 Notes Exercises9 Optimization 9.1 Reachability and decomposibility 9.2 Linear programming formulations 9.3 Multiobjective optimization 9.4 Optimality equations 9.5 Algorithms 9.6 Optimization in networks 9.7 One-dimensional inventory model 9.8 Hedging and workload 9.9 Notes Exercises10 ODE Methods 10.1 Examples 10.2 Mathematical preliminaries 10.3 Fluid limit model 10.4 Fluid-scale stability 10.5 Safety stocks and trajectory tracking 10.6 Fluid-scale asymptotic optimality 10.7 Brownian workload model 10.8 Notes Exercises11 Simulation and Learning 11.1 Deciding when to stop 11.2 Asymptotic theory for Markov models 11.3 The single-server queue 11.4 Control variates and shadow functions 11.5 Estimating a value function 11.6 Notes ExercisesAppendix Markov ModelsBibliographyIndex
插图：More history on Brownian models is contained in the Notes for Chapter 5. This book provides foundations for resource allocation and perfonnance evaluation, but can- not go too deeply into specific issues in each possible application. A notable example is the area of Internet congestion control where there are many constraints due to the reliance on ar- chitecture and algorithms designed in the 1970s. Srikant's monograph [459] treats this problem in-depth using a range of techniques, including variants of methods described in this book. Although much of this book concerns the construction and analysis of algorithms to con- struct feedback laws for control, to bound performance, or to improve simulation, this book does not contain any theory of algorithms. In particular, we do not touch upon complexity the- ory for algorithms as described in [390, 391,392, 115, 42, 194], although this theory is the most important motivation for the approximation techniques developed in the book. The optimal control problems posed in this book are primarily centralized in the sense that there is a centralized decision maker that possesses complete information. A decentralized con- tro[ solution is one that can be implemented based on local information, such as nearby con- gested links. For a physical network such as the Internet, or the North American power grid, a centralized control framework is absurd. For example, in a power distribution system generators may be owned by different companies, who supply power to various utilities, using power lines man- aged by different system operators. Methods from game theory can be applied to study the consequences of potential outcomes in a decentralized noncooperative setting [31,412]. We do not address any of these game-theoretic issues. However, the centralized optimal policy can be used as a benchmark against which the performance of a decentralized system is evaluated.Moreover, we do consider classes of policies that can be implemented using only local in- formation. One example is the class of MaxWeight policies introduced in Section 4.8. These are a subset of myopic: policies. In some cases it can be shown that a myopic policy is approx- imately optimal if the network is congested, or the network load is high （see Chapter 9 and Theorem 10.0.2）.
无
跟现在复杂网络研究的主流方向好像有点不一样吧，再看看吧
不愧是IEEE follow的书，写的挺好！
正在学习中，希望能学到一点东西
四天以内送达的
经典之作，满意
比较全面，有助于学习专业英文