实时移动云计算：新兴技术的推动

需积分: 2 199 浏览量更新于2024-07-19 收藏 31.35MB PDF 举报

《实时移动云计算通过新兴技术实现》是一本由托尔加·索亚塔（Tolga Soyata）撰写的专著，收录在无线技术和电信进步系列（Advances in Wireless Technologies and Telecommunication, AWTT）中。该书由信息科学参考（IGI Global）在美国出版，版权归属2015年IGI全球。本书探讨了如何利用前沿科技推动实时移动云计算的发展，这在当今移动互联网时代具有重要的实践意义。章节内容涵盖了实时移动云计算的关键要素，如云计算的基本概念、移动设备的性能需求、边缘计算、5G网络、物联网（IoT）、大数据分析、人工智能（AI）以及区块链等新兴技术在提升移动应用响应速度、降低延迟和优化资源利用率方面的作用。作者可能对这些技术在实时数据处理、实时决策支持、实时通信服务等方面的集成应用进行了深入剖析。书中可能强调了如何解决移动设备的电池寿命、网络连接不稳定性和数据安全等问题，同时探讨了如何通过云计算平台提供无缝的用户体验，尤其是在低带宽和高并发场景下。对于企业而言，这本书可能提供了关于如何利用这些技术构建可扩展、高性能的移动应用策略的指导。由于版权原因，未经许可，任何复制、存储或分发行为都必须得到出版商的书面授权。此外，书中可能会包含案例研究和实证分析，以展示在不同行业和应用场景中如何成功地将新兴技术与实时移动云计算相结合，实现业务价值。为了获取更详细的信息，读者可以访问美国亚马逊官网查找该书，或者直接联系出版社获取原著的电子版或实体书。图书馆馆藏目录和英国图书馆的Cataloguing in Publication 数据也可能提供了额外的索引和引用信息。《实时移动云计算通过新兴技术实现》是一本极具价值的专业书籍，对于IT专业人士、移动应用开发者、网络工程师和战略规划者来说，它提供了一个深入理解并应用于实时移动云计算领域的宝贵资源。

Preface

Despite such unsuccessful attempts as the external GPS functionality I just mentioned, the introduc-

tion of the PDAs and added functionality inside the phones sparked an avalanche of interest within tech

enthusiasts and business folks that wanted to use these devices for business functions: as a calculator,

email device, storing contacts, or whatever else you could introduce at a reasonable price. To be able

to perform these functions, Palm and Blackberry had to incorporate a Radio Processor (or Baseband

Processor) and an Application Processor into these devices. The Radio Processor was responsible for

the phone call functionality, implementing the previously mentioned sophisticated data encoding and

communication. Application Processor, on the other hand, was responsible for turning the phone into a

computer (almost). It was only a matter of time when the Application Processor became sophisticated

enough to turn this device into a full-blown computer, and was eventually accompanied with his sister:

Media Processor, which was responsible for heavy-duty signal or image processing. For this to materialize,

a lot of simultaneous progress was needed in different fields: VLSI technology had to advance to a point

where hundreds of millions of transistors could be built into the Application and Media Processor chips,

and, major advances in Computer Architecture were needed to make sure that, this device could operate

at a very low (1W or 2W) power budget, while delivering the computation that these applications needed.

Battery technology didn’t advance as fast as the previous two, but a deeper understanding of the Lithium

Ion rechargeable batteries a allowed more intelligent usage of them, which in turn increased battery life.

The term smart phone originates in this era, when mobile devices could perform so many different

functions that, they even started communicating with their user to improve Human-Computer Interaction

(e.g., the Siri on the iPhone and many similar implementations in other brands).We are now in an era

where these smart phone devices are an indispensible part of human life, connecting us to the internet

and social networks. The breakneck speed in application development put almost every imaginable

application in the market which can be inexpensively purchased and run on smart phones and the next

natural question is: what do we go from here ? The previous half decade has seen an explosion of research

interest in answering this question. With very stable and fast connections to the internet backbone, smart

phones’ capabilities were no longer limited to their own hardware. One parallel development effort aimed

to take advantage of an emerging concept: the cloud.

When the internet connection speeds of smart phones reached a threshold and became increasingly

more affordable through the introduction of faster data connection standards such as 3G, 4G, and LTE,

it became possible to augment the capabilities of smart phones with the vast resources residing in large

scale datacenters (the cloud). This synergistic coupling, Mobile-cloud computing, marked a new era

in the development for smart phone applications. Using Mobile-Cloud Computing allowed using less

capable smart phones to perform highly sophisticated functions, partly making the capabilities of the

smart phone itself less relevant. Additionally, offloading parts (or all) of a mobile application to the cloud

could save precious battery life. More excitingly, Mobile-Cloud Computing could allow smart phones

to run applications that they could never run themselves in the foreseeable future, due to the limitation

of their resident hardware.

Mobile Cloud Computing is in its infancy, much like the smart phone itself was a decade ago. Much

research effort will be devoted to making it a usable computational and resource sharing model in the

following decade. There will be missteps and major success stories. One thing that is for sure: its continu-

ous progress will never stop. Using Mobile-Cloud Computing, combined with the future communication

standards such as LTE Advanced and 5G that aim much higher data rates than what is available today,

it will be possible to run applications that will never be possible to run on mobile devices alone. Such

applications are extremely resource intensive (computation, memory, and storage-wise) and may require

Preface

access to real-time data that is only resident in the cloud. One such family of applications, ones that

utilize Real-time Mobile Cloud Computing, is the focus of this book. A good representative example

of this application family is Real-time Mobile-Cloud Face Recognition, which initially motivated the

authoring of this book. The family of resource-hungry mobile applications are not designed to extend

mobile applications to the cloud, but rather, they are designed as mobile-cloud applications right from

the start, since it is not possible to run them solely on mobile devices.

This book contains a total of 12 tutorial and research chapters which describe new and innovative

mobile applications and provide supporting surveys to understand the operating characteristics of these

applications. The applications described in this book are resource-hungry in different ways: Some of them

require an extensive amount of processing power, RAM (short-term memory) or flash/hard disk storage

(long-term memory). Some require access to an enormous amount of data that is updated in real-time.

Such a quantity of data (e.g., Peta bytes) is far beyond the capability of any single device to process or

handle, including smart phones (i.e., Big Data). All of these challenges introduce many research direc-

tions to make these exciting applications a reality. Solutions to these challenges span multiple disciplines

in Electrical Engineering and Computer Science, and is the primary focus of this book. The chapters of

this book are organized as follows:

Chapter 1 describes a concept system for remote health monitoring of cardiac patients outside a

healthcare organization. Four separate components of this system are described in detail: The bio-sensor

component of the system involves the design of custom advanced sensors that are capable of detecting

proteins such as Troponin, Myoglobin and CRP, that are essential for advanced cardiac patient moni-

toring. The custom circuit interface for these sensors is detailed and the operational characteristics are

described. Communication interface of the system uses standardized communication components found

in mobile cloud computing (e.g., cloudlet), while an integration to the emerging Internet-of-Things (IoT)

infrastructure is also described in terms of the concentrator-cloudlet co-operation. Finally, the authors aim

at formulating a new visualization mechanism for cardiac data that can provide summarized information

over the duration of the long-term health monitoring. This is aimed at remedying the shortcomings of

the short-term, in-hospital ECG recordings that only provide limited information about the evolution of

the patients’ health condition.

Chapter 2 provides an extensive survey of Military Tactical Communications and Networking

(MTCAN). To summarize, MTCAN algorithms and protocols are similar to Mobile Ad Hoc networks

(MANETs), however, they are distributed to introduce robustness and scalability and to eliminate single

points of failure. Authors provide a survey of the TRACE (Time Reservations using Adaptive Control for

Energy efficiency) family of protocols that have been developed at the University of Rochester, through

the financial support of Harris Corporation, RF Communications Division, which is a leader in MTCAN.

The protocols that are surveyed in the chapter include SH-TRACE (Single-Hop Time Reservation using

Adaptive Control for Energy efficiency), MH-TRACE (Multi Hop Time Reservation using Adaptive

Control for Energy efficiency), NB-TRACE (Network-wide Broadcasting through Time Reservation

using Adaptive Control for Energy efficiency), MC-TRACE (Multi Casting through Time Reservation

using Adaptive Control for Energy efficiency), MMC-TRACE (multi-rate multicasting), AR-TRACE

(adaptive redundancy), CDCA-TRACE (Cooperative node balancing and dynamic channel allocation

with Time Reservation using Adaptive Control for Energy efficiency) and U-TRACE (Unified TRACE).

the authors conclude by providing future research directions.

xvi

Preface

Chapter 3 provides a survey of the technologies, infrastructure components, and research leading

to the formation of an emerging cloud-based service: Sensing as a Service (S

aaS). In this new service

offering, any mobile device that is connected to the internet can be a part of the sensing network. Wire-

less Sensor Networks (WSNs), mobile phones, and other emerging Internet-of-Things (IoT) devices can

contribute their local data to an application that aggregates this information with the intent to provide a

much more comprehensive version of this data (i.e., a globalized or much more expanded version). This

data can be used to provide statistics or sensing results to all (or some) of the contributing nodes. Since

the only limitation for the mobile devices is a form of connection to the sensing network, and, clearly, the

willingness of a node to contribute all of part of the data, the sensing network could be extremely flex-

ible. Finally, the global authority for the application is in the cloud with almost no resource constraints,

especially as compared to the contributing small nodes. This could be used to perform resource-intensive

analytics on the received data, thereby opening the door to a set of exciting applications and service

offerings. However, this form of a sensing network introduces significant challenges such as privacy

concerns and reliability of the acquired data. This chapter provides a survey of the techniques to deal

with such challenges.

Chapter 4 introduces a remote health monitoring system where a patient’s vitals are being monitored

in his/her house and the results are being transmitted to the cloud. An algorithm running in the cloud

detects potential hazardous health conditions, such as abnormal cardiac function and warns the doctor

in real-time. During the execution of this algorithm, the acquired patient medical data is transmitted

from the patient’s house into the cloud and it is processed in the cloud. The results are transmitted into

the doctor’s tablet. This chapter primarily focuses on the privacy of the data during the transmission.

An emerging encryption technique, called Fully Homomorphic Encryption (FHE), is used to encrypt

the data that is being transmitted. However, this type of encryption causes significant expansion of the

data and is highly computationally-intensive to process. Traditionally, circuit-based solutions are used in

conjunction with FHE by turning each computation into a circuit. This circuit is implemented using FHE

building blocks to yield the final result. A different approach is introduced in this chapter which uses a

Branching Program. In their approach, each Yes/No decision (such as a patient has a health hazard vs.

does not have a health hazard) is formulated as a Branching Program, which has a True/False answer.

This Branching Program can be solved using FHE building blocks. While this limits the applicability of

FHE to a restricted set of medical computations, authors report a 20x speed-up in the overall execution

time for certain useful medical applications.

Chapter 5 introduces a computational infrastructure where multiple smart phones contribute their

computational power to an application. This concept, termed Volunteer Computing, allows computation-

ally-intensive applications to be run by a set of distributed mobile devices that are owned by volunteers

that opt-in to the computational network. Authors break down the approaches into major categories

that are defined by the volunteering entity: 1) Server-driven mobile-distributed computing approaches

surveyed in this chapter are a) prime-number research, b) text search; π value estimation, c) distributed

hash-cracking, d) distributed multimedia search, e) dynamic life-cycle assessment of a building, and f)

DNA sequence matching. 2) User-driven approaches are a) speech-to-text, b) face detection and collab-

orative photography, c) collaborative file download, and d) testing of the cloudlet formation algorithm.

3) Mobile Volunteer Computing approaches include a) scientific research projects, b) protein structure

predictions and multiple other ones. The authors conclude with a summary and provide pointers to

future research directions.

xvii

Preface

Chapter 6 details a new cloud-based service model called Acceleration as a Service (AXaaS). Authors

base the conceptualization of this new aaS model on the acceleration capability of a cloudlet. To speed-up

mobile-cloud applications, such as real-time face recognition, it is known that, a cloudlet can be placed

locally within the WiFi reach of a mobile device, providing a fast and dedicated link to the cloud. The

significant difference between a traditional cloudlet and what is prescribed in this chapter is where the

cloudlet is placed. Rather than the standard model where a user owns a cloudlet, renting the cloudlet

is suggested. The best entity for renting the cloudlet is determined to be the Telecom Service Provider

(TSP). The TSP can rent a cloudlet through a monthly fee, which eliminates the need for the user to

own or upgrade such a device. Since it is very expensive to shuttle data back and forth between mobile

devices and the TSP, one important research challenge is to determine which parts of the code require

computational speed-up (i.e., acceleration). The significant research challenge in such an infrastructure

is what to accelerate using the TSP. This is analyzed by the authors through code profiling. It is deter-

mined that, a very limited library of functions, such as Fast Fourier Transform (FFT) and Basic Linear

Algebra Subroutines (BLAS) are sufficient as acceleration points. Merely accelerating these two API

functions provides drastic overall application speed-up, thereby creating a TSP service that the users are

willing to pay for. A business ROI analysis is also provided from the standpoint of the TSP and the user.

Chapter 7 provides a framework for generalized privacy-preserving medical cloud computing.

Running medical applications in the cloud offers a healthcare organization significant cost savings by

being able to outsource the storage and computation of medical data to a cloud operator. However, this

implies that, the breach of the medical data during cloud computing could result in a violation of Health

Information Portability and Accountability Act (HIPAA) and is not acceptable. Authors propose to use

Fully Homomorphic Encryption (FHE) to operate on the medical data that is stored in the cloud. While

the storage of data can preserve privacy by using traditional encryption algorithms such as Advanced

Encryption Standard (AES), no computation can be done on AES-encrypted data. On the other hand,

FHE-based encryption allows both storage and computation in a privacy-preserving fashion, since the

cloud cannot observe the data that it is computing. However, FHE-based computation and storage are

substantially more costly than their AES-counterpart. Authors formulate a mechanism where some non-

computationally-expensive medical computations can be performed in the cloud, such as minimum/

maximum/average heart rate computations during remote patient monitoring, as well as the detection

of certain cardiac health hazards such as Long-QT syndrome (LQTS). A detailed analysis is provided

for each step of these computations.

Chapter 8 provides a survey of hardware and software support for Virtual-Machine (VM) based

mobile-cloud application offloading. Mobile offloading concepts are briefly surveyed and the areas where

VMs can help are determined. Advantages and disadvantages of virtualization are explained in detail.

This is followed by a survey of software support including virtualization approaches, hypervisor types,

Operating System support, as well as existing VM types. A survey of hardware assisted virtualization is

also provided which details the Instruction Set support that is resident in three popular manufacturers’

CPUs: Intel, AMD and ARM. Hardware assisted virtualization allows the virtualization of memory,

I/O, and the CPU cores. Challenges in virtualization are listed such as the security and overhead of

virtualization. A summarized discussion of GPU virtualization is also provided.

Chapter 9 is a practical and theoretical tutorial on network latency measurement. The author provides

an extensive set of latency measurements provided by his research team. These measurements are done in

a PlanetLab environment which is a network of volunteering academic institutions. A detailed discussion

of how to repeat these experiments is also provided through a brief introduction to PlanetLab. Based on

xviii

Preface

the measured data, author draws conclusions about what the distributions of the global latency values

are. Latency is an important performance-limiting factor for mobile-cloud applications as it determines

the speed at which a mobile device can access the cloud. The intuitions and practical measurements pro-

vided in this chapter are, therefore, an important guide to designing real-time mobile-cloud applications.

Chapter 10 focuses on the business aspects of running highly resource-intensive applications in the

cloud and the cost of outsourcing such applications into the cloud. A list of the operating costs for the

three popular cloud operators (Amazon, Google, and Microsoft Azure) are provided. These operators

charge fees based on a) storage, b) computation, and c) network traffic. The analysis provided in this

chapter breaks the application costs into these three categories and associates metrics for each one of

these parameters. This analysis is not trivial in that, cloud operators provide a vast array of options based

on commitment levels (in terms of time duration). Authors suggest the best choices for these options

based on the characteristics of two resource-intensive applications: 1) a remote health monitoring ap-

plication using Fully Homomorphic Encryption to perform privacy-preserving medical cloud computing,

and 2) real-time mobile-cloud face recognition. Both of these applications are analyzed in terms of their

functional characteristics and their resource demands are determined. These results are converted to a

monthly operating cost in the cloud for three different cloud operators.

Chapter 11 provides a technical and practical tutorial for understanding and running face recognition,

which is a good representative case for resource-intensive mobile-cloud applications. The theory behind

this application is described by breaking the application into three of its distinct execution phases: Face

Detection, Projection, and Search. The theory behind these three different phases is detailed and the steps

required to run this application on commodity computers using the OpenCV library are provided in a

tutorial format. By following the steps that are provided in this chapter, other researchers in resource-

intensive mobile-cloud computing can easily run face recognition using freely available open source tools.

Chapter 12 provides a practical view to the expected characteristics of mobile platforms. The author

is a retired USG officer who brings a practical view to the importance of these parameters in general-

ized radio communication and draws conclusions from his practical experience. Availability (Ao) is

explained as one of the most important parameters and the details are provided in regard to Ao and its

effect on radio communication.

Tolga Soyata

University of Rochester, USA

xix

剩余420页未读，继续阅读

yinkaisheng-nj

粉丝: 762
资源: 6231

实时移动云计算：新兴技术的推动

Advances in Big Data and Cloud Computing 无水印原版pdf

Research Advances in Cloud Computing 无水印原版pdf

Intelligent Cloud Computing 无水印原版pdf

SubTrack: Enabling Real-Time Tracking of Subway Riding on Mobile Devices

key-5g-physical-layer-technologies-enabling-mobile-and-fixed-wir

Compact and stable real-time dual-wavelength digital holographic microscopy with a longworking distance objective

C-band real-time 400/300 Gb/s OOK bidirectional interconnection over 20 km multicore fibers

2009中华医院信息网络大会:TOWARDS THE FUTURE OF NURSING Enabling Nurses to Use Enabling Technologies-marion ball

藏经阁-SparkOscope--Enabling Apache Spark Optimization through Cros

VMworld 2009 - DC32: Technologies Enabling The Next Generation of Cloud Services

最新资源