English
Big Data and Computational Intelligence in Networking-CRC(2018).pdf
文件大小：
12437k
资源说明：Recent years have witnessed a deluge of network data propelled by the emerging online social media, user-generated video contents, and global-scale communi- cations, bringing people into the era of big data. Such network big data holds much critical and valuable information including customer experiences, user behaviors, service levels, and other contents, which could significantly improve the efficiency, effectiveness, and intelligence on the optimization of the current Internet, facilitate the smart network operation and management, and help service providers and content providers reduce capital expenditure (CapEx) and opera- tional expenditure (OpEx) while maintaining a relatively high-level quality of service (QoS) and quality of experience (QoE).
Typical examples of network intelligence received from network big data include rapid QoE impairment detection and mitigation, optimization of network asset utilization, proactive maintenance, rapid outage restoration, and graceful disaster recovery. These aims can be achieved from high-level computational intelligence based on emerging analytical techniques such as big data pro- cessing, Web analytics, and network analytics employing software tools from advanced analytics disciplines such as machine learning, data mining, and pre- dictive analytics. The computational intelligence for big data analysis is playing an ever-increasingly important role in supporting the evolution of the current Internet toward the next-generation intelligent Internet.
However, the unstructured, heterogeneous, sheer volume and complex nature of network big data pose great challenges on the computational intelligence of these emerging analytical techniques due to high computational overhead and communication cost, non-real-time response, sparse matrix-vector multi- plications, and high convergence time. It is therefore of critical importance to understand network big data and design novel solutions of computational intelligence, scaling up for big data analytics of large-scale networks to auto- matically discover the hidden and valuable information available for smart network operations, management, and optimization. This has been established as
ix
x Preface
a new cross-discipline research topic in computer science, requiring anticipation of technical and practical challenges faced by mixed methods across multiple disciplines.
In this book, we have invited world experts in this area to contribute the chapters that cover the following four parts:
1. Part1:BasicsofNetworkedBigData:Thisparthelpsunderstandtheprop- erties, characteristics, challenges, and opportunities of networked big data, geospatial data, and wireless big data. This part covers the following:
a. Mathematical properties: A variety of aspects related to networks, including their topological and dynamical properties, as well as their applications to real-world examples
b. Geospatial data and geospatial semantic web: Challenges and opportunities of the geospatial semantic web brought for sharing and utilizing big geospatial data
c. Big data over wireless networks: Typical scenarios, various chal- lenges, and potential solutions for wireless transmission of big data
2. Part 2: Network Architecture for Big Data Transmissions: This part presents new proposals and network architectures to ensure efficient big data transmissions and streaming big data processing.
a. Big data transfer: Challenges of bandwidth reservation service for efficient big data transfer and the potential solutions
b. Internet of Things (IoT): A dynamic and independent Cloud com- puting architecture based on a service-oriented architecture for IoT devices, to allow users to freely transfer their IoT devices from one vendor to another
c. Streamingbigdataprocessing:HowtomaximizeQoSandminimize OpEx when performing task scheduling and resource allocation in geo-distributed Clouds
3. Part3:AnalysisandProcessingofNetworkedBigData:Thispartexplains how to perform big data analytics based on emerging analytical techniques such as big data analytics, Web analytics, network analytics, and advanced analytics disciplines such as machine learning, data mining, and predictive analytics. This part covers the following areas:
a. Alternatingdirectionmethodofmultiplier(ADMM):Itsapplications to large-scale network optimizations
b. Dynamicnetworkmanagementandoptimization:Rethinkofcurrent network analysis, management and operation practices; impact of
Preface xi network evolution on the computation of key network metrics;
hyperbolic big data analytics
c. Predictiveanalyticsandsmartretrieval:Utilizethenetworkbigdata by performing a data, information, knowledge, and wisdom (DIKW) hierarchy to the product of its processes
d. Recommendation systems: Key challenges and solutions for data sparsity problem, data scale issue, and cold-start problem
e. Coordinate gradient descent methods: Unconstrained convex mini- mization problems with differentiable objective function in network problems
f. MapReduce: Data locality and dependency analysis; dependency- aware locality for MapReduce
g. Distributed machine learning: Big data and big models for network big data; how to parallelize parameter updates on multiple work- ers; how to synchronize concurrent parameter updates performed by multiple workers
h. Biggraph:Biggraphdecomposition;real-timeandlarge-scalegraph processing; big data security
4. Part 4: Emerging Applications of Networked Big Data: This part covers some emerging applications on the following:
a. Intelligent mall shopping: Location-based mobile augmented real- ity applications; using network data to enable intelligent shopping; robust feature learning in cold-start heterogeneous-device localiza- tion; learning to query in the cold-start retailer content
b. Networkanomalydetection:Howtoefficientlyusenetworkbigdata to perform accurate anomaly detection
c. Transportation: Advances of spatial network big data (SNBD) tech- niques; challenges posed by SNBD in transportation applications and the potential solutions
d. Biomedical and social media domain: Graph as a representation schema for big data; graph-based models and analyses in social text mining, and bioinformatics and biomedical
e. Smart manufacturing: Big data characteristics in manufacturing; data collection and data mining in manufacturing; applications of big data in manufacturing.
This book presents the state-of-the-art solutions to the theoretical and prac- tical challenges stemming from the leverage of big data and its computational intelligence in supporting smart network operation, management, and optimiza- tion. In particular, the technical focus covers the comprehensive understanding
xii Preface
of network big data, efficient collection and management of network big data, distributed and scalable online analytics for network big data, and emerging applications of network big data for computational intelligence.
Targeted audiences: This book targets both academia and industry readers. Grad- uate students can select promising research topics from this book that are suitable for their thesis or dissertation research. Researchers will have a deep under- standing of the challenging issues and opportunities of network big data and can thus easily find an unsolved research problem to pursue. Industry engineers from IT companies, service providers, content providers, network operators, and equipment manufacturers can get to know the engineering design issues and cor- responding solutions after reading some practical schemes described in some chapters.
We have required all chapter authors to provide as much technical detail as possible. Each chapter also includes references for readers’ further studies and investigations. If you have any comments or questions on certain chapters, please contact the chapter authors for more information.
Thank you for reading this book. We wish that this book will help you with the scientific research and practical problems of network big data.
本源码包内暂不包含可直接显示的源代码文件，请下载源码包。
