PDF Fog Radio Access Networks F RAN Architectures Technologies and Applications Mugen Peng Download

Download and Read online, DOWNLOAD EBOOK, [PDF EBOOK EPUB ], Ebooks
download, Read Ebook EPUB/KINDE, Download Book Format PDF
Fog Radio Access Networks F RAN Architectures

Technologies and Applications Mugen Peng
OR CLICK LINK
https://textbookfull.com/product/fog-radio-access-
networks-f-ran-architectures-technologies-and-
applications-mugen-peng/
Read with Our Free App Audiobook Free Format PFD EBook, Ebooks dowload PDF
with Andible trial, Real book, online, KINDLE , Download[PDF] and Read and Read
Read book Format PDF Ebook, Dowload online, Read book Format PDF Ebook,
[PDF] and Real ONLINE Dowload [PDF] and Real ONLINE
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
5G radio access networks : centralized RAN, cloud-RAN,

and virtualization of small cells 1st Edition Trestian
https://textbookfull.com/product/5g-radio-access-networks-
centralized-ran-cloud-ran-and-virtualization-of-small-cells-1st-
edition-trestian/
Blockchain-enabled Fog and Edge Computing: Concepts,

Architectures and Applications: Concepts, Architectures
and Applications 1st Edition Muhammad Maaz Rehan
(Editor)
https://textbookfull.com/product/blockchain-enabled-fog-and-edge-
computing-concepts-architectures-and-applications-concepts-
architectures-and-applications-1st-edition-muhammad-maaz-rehan-
editor/
Blockchain-enabled Fog and Edge Computing: Concepts,

Architectures and Applications-Concepts, Architectures
and Applications 1st Edition Muhammad Maaz Rehan
(Editor)
https://textbookfull.com/product/blockchain-enabled-fog-and-edge-
computing-concepts-architectures-and-applications-concepts-
architectures-and-applications-1st-edition-muhammad-maaz-rehan-
editor-2/
Spectrum Access and Management for Cognitive Radio

Networks 1st Edition Mohammad A Matin (Eds.)
https://textbookfull.com/product/spectrum-access-and-management-
for-cognitive-radio-networks-1st-edition-mohammad-a-matin-eds/
Photonic Applications For Radio Systems Networks Fabio
Cavaliere
https://textbookfull.com/product/photonic-applications-for-radio-
systems-networks-fabio-cavaliere/
Elastic Optical Networks Architectures Technologies and

Control 1st Edition Víctor López
https://textbookfull.com/product/elastic-optical-networks-
architectures-technologies-and-control-1st-edition-victor-lopez/
Privacy Enhancing Fog Computing and Its Applications

Xiaodong Lin
https://textbookfull.com/product/privacy-enhancing-fog-computing-
and-its-applications-xiaodong-lin/
Media Convergence and the Development Strategies of

Radio and Television in China 1st Edition Peng Duan
https://textbookfull.com/product/media-convergence-and-the-
development-strategies-of-radio-and-television-in-china-1st-
edition-peng-duan/
Media Convergence and the Development Strategies of

Radio and Television in China 1st Edition Peng Duan
https://textbookfull.com/product/media-convergence-and-the-
development-strategies-of-radio-and-television-in-china-1st-
edition-peng-duan-2/
Wireless Networks
Mugen Peng
Zhongyuan Zhao
Yaohua Sun
Fog Radio
Access
Networks
(F-RAN)
Architectures, Technologies, and
Applications
Wireless Networks
Series Editor
Xuemin Sherman Shen
University of Waterloo
Waterloo, ON, Canada
The purpose of Springer’s new Wireless Networks book series is to establish
the state of the art and set the course for future research and development in
wireless communication networks. The scope of this series includes not only all
aspects of wireless networks (including cellular networks, WiFi, sensor networks,
and vehicular networks), but related areas such as cloud computing and big data.
The series serves as a central source of references for wireless networks research
and development. It aims to publish thorough and cohesive overviews on specific
topics in wireless networks, as well as works that are larger in scope than survey
articles and that contain more detailed background information. The series also
provides coverage of advanced and timely topics worthy of monographs, contributed
volumes, textbooks and handbooks.
More information about this series at http://www.springer.com/series/14180

Mugen Peng • Zhongyuan Zhao • Yaohua Sun
Fog Radio Access Networks

(F-RAN)
Architectures, Technologies, and Applications
Mugen Peng Zhongyuan Zhao
School of Information and Communication School of Information and Communication
Engineering Engineering
Beijing University of Posts and Beijing University of Posts and
Telecommunications Telecommunications
Beijing, China Beijing, China
Yaohua Sun
School of Information and Communication
Engineering
Beijing University of Posts and
Telecommunications
Beijing, China
ISSN 2366-1186 ISSN 2366-1445 (electronic)

Wireless Networks
ISBN 978-3-030-50734-3 ISBN 978-3-030-50735-0 (eBook)
https://doi.org/10.1007/978-3-030-50735-0
© Springer Nature Switzerland AG 2020

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.
This Springer imprint is published by the registered company Springer Nature Switzerland AG.
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface
In the fifth generation (5G) radio access networks and beyond, a paradigm of
fog computing-based radio access network (F-RAN) has emerged to meet the
requirements of the explosively increasing high-speed applications and the massive
number of Internet-of-Things (IoT) devices. Inherited from both heterogeneous
networks (HetNets) and cloud computing-based radio access networks (C-RANs),
F-RANs take full advantages of cloud computing, fog computing, and heterogenous
cooperative processing. F-RANs can coordinate the severe interference via the adap-
tive distributed and centralized processing techniques and provide great flexibility
to satisfy quality-of-service requirements of various intelligent applications and
services, such as the ultralow-latency of about 1.0 ms required by self-driving based
Internet of vehicles, and up to 99.999% reliability, and even 106 connections/km2
density as required by intelligent manufacturing.
Since F-RAN first combines fog computing, cloud computing, and HetNets
together and has been regarded as the key evolution path to the 5G beyond
system, it has drawn a large number of attention from both academia and industry.
As integrating with artificial intelligence (AI), non-orthogonal multiple access
(NOMA), and other advanced emerging technologies, F-RANs have entered a new
researching and development era to solve the challenges that 5G beyond system
meets. Actually, it is well known that F-RAN is potentially an evolutionary path
to the sixth generation (6G) mobile system. This is a cutting-edge technique of
multiple disciplines, including AI, wireless networks, radio signal processing, fog
computing, and cloud computing. The versions and classical application scenarios
defined in 5G are hardly fulfilled in 2020, while they will be fully provided in 6G.
In particular, 6G can meet requirements of enhanced mobile broadband, massive
machine-type communications, and ultra-reliable and low-latency communications
for rich IoT services, which will penetrate various industry and business applica-
tions. In terms of these rich IoT services, F-RAN can provide a unified framework
for massive access of heterogonous IoT devices, which simplifies the control
and management mechanisms. With respect to AI applied in F-RANs, sufficient
computation resources in fog nodes can be provided to execute machine learning
v
vi Preface
and deep learning algorithms, which means that both the cost and the training
efficiencies can be significantly improved.
As the F-RAN moves from the theoretical research to real world applications
industry and academia are working together towards the protocols defined in
standards and algorithms of all air interface layers in products, so as to enable
spectral-, energy-, and cost-efficient F-RANs to be widely used as a key solution
for 5G beyond and even 6G systems.
This book is firstly intended to present a comprehensive overview framework
of recent advances in F-RANs, from both the academia and industry perspectives.
In particular, this book covers the architecture, performance analysis, physical-
layer design, resource allocation, computation offload, and field trials. The recent
academic research results of F-RANs, such as the analytical results of theo-
retical performance limits and the optimization theory-based resource allocation
algorithms, have been introduced. Meanwhile, to promote the implementation of F-
RANs, the latest standardization procedure and testbed design have been discussed
as well. Finally, this book will be concluded by summarizing the existing open issues
and future trends of F-RANs.
We sincerely hope that this book will serve as a powerful reference for engineers
and students in the majors of electronic, computations, and communications, which
results in motivating a large number of students and researchers to tackle these
numerous open issues and challenges highlighted in F-RANs for 5G beyond and
even 6G systems.
Beijing, China Mugen Peng

Beijing, China Zhongyuan Zhao
Beijing, China Yaohua Sun
Acknowledgements
First, we would like to express our sincere gratitude to all contributors, without
whom this book would have never been published. Our contributors are as follows:
Mugen Peng, Zhongyuan Zhao, Yaohua Sun, Shi Yan, Chenxi Liu, Bin Cao, Xiqing
Liu, Hongyu Xiang, Kecheng Zhang, Binghong Liu, Zhendong Mao, Wenbin Wu,
Tian Dang, Ling Qi, Yangcheng Zhou, Yuan Ai, Xinran Zhang, Xian Zhang, Bonan
Yin, and Wenyun Chen. Meanwhile, we would like to thank Springer Press staff for
their continuous encouragement and support.
Second, we would like to thank our colleagues and friends all over the world who
promoted fog computing into cellular networks and Internet-of-Things (IoT) for 5G
and 6G developments. This book was supported in part by the State Major Science
and Technology Special Project (Grant No. 2018ZX03001025-004), the National
Natural Science Foundation of China under No. 61925101, 61921003, 61831002,
and 61901044, the Beijing Natural Science Foundation under No. JQ18016, and
the National Program for Special Support of Eminent Professionals. Without the
support of these funding, F-RANs are hard to formulate and the corresponding
research works could be hardly achieved within a short time. Meanwhile, I would
like to thank Prof. H. Vincent Poor and Dr. Chonggang Wang because they helped
open the window of C-RANs when Prof. Mugen Peng was an academic visiting
fellow at Princeton University.
Third, we also thank all of those who have played a part in the preparation of
this book. Our coordinator, Hemalatha Velarasu, provided useful guidelines and
monitored the whole process carefully. Our Senior Editor, Mary E. James, arranged
the book review and fed back the review comments in a very short time. Also, special
thanks are given to Production Editors, Brian Halm and Brinda Megasyamalan, as
well as Assistant Editor, Zoe Kennedy.
Most importantly, we appreciate Series Editor, Prof. Xuemin (Sherman) Shen,
for his great support and invaluable suggestions.
Last but not least, we would also like to sincerely thank the support of our family
members for supporting our research work when we were away from home.
vii
Contents
1 Brief Introduction of Fog Radio Access Networks . . . . . . . . . . . . . . . . . . . . . . 1

1.1 History and Evolution of RANs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1.1 HetNet Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.1.2 C-RAN Overview. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.1.3 5G Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.1.4 6G Prospect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.2 Fog Computing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.2.1 Fog Network Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.2.2 Fog and Edge Computing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.3 F-RANs for 5G and Beyond . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.4 Relative Standards for F-RANs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.5 Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2 System Architecture of Fog Radio Access Networks . . . . . . . . . . . . . . . . . . . 21
2.1 System Architecture of F-RANs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.2 Network Slicing in F-RANs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.3 Radio Interface Techniques in F-RANs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.4 Application Cases for F-RANs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.4.1 F-RAN Enabled V2X. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.4.2 F-RAN Enabled AR/VR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.4.3 F-RAN Enabled Space Communications . . . . . . . . . . . . . . . . . . 31
2.4.4 F-RAN Enabled UAV Communications . . . . . . . . . . . . . . . . . . . 33
2.5 F-RAN Evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
2.5.1 System Architecture for AI-driven F-RANs . . . . . . . . . . . . . . 35
2.5.2 Principles of AI in F-RANs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.5.3 Challenges and Future Works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.6 Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
ix
x Contents
3 Theoretical Performance Analysis of Fog Radio Access Networks . . . 41

3.1 Ergodic Capacity of User Association in F-RANs . . . . . . . . . . . . . . . . . 41
3.1.1 System Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.1.2 Performance Analysis of Ergodic Capacity . . . . . . . . . . . . . . . 42
3.1.3 Numerical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.2 Effective Capacity with Content Caching in F-RANs . . . . . . . . . . . . . . 49
3.2.1 System Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.2.2 Performance Analysis of Effective Capacity . . . . . . . . . . . . . . 52
3.2.3 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
3.3 Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4 Cooperative Signal Processing in Fog Radio Access Networks . . . . . . . . 61
4.1 Cooperative Non-orthogonal Multiple Access in F-RANs . . . . . . . . . 61
4.1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.1.2 System Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.1.3 Performance Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.1.4 Performance Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.2 Contract Based Cooperative Signal Processing . . . . . . . . . . . . . . . . . . . . . 68
4.2.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
4.2.2 System Model and Cooperative Signal Processing
Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.2.3 Optimal Contract Design Under Complete CSIs . . . . . . . . . . 72
4.2.4 Contract Design Under Practical Channel Estimation . . . . 75
4.2.5 Numerical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
4.3 Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
5 Flexible Network Management in Fog Radio Access Networks . . . . . . . 85
5.1 The Access Slicing Paradigm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
5.1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
5.1.2 A Hierarchical Management Architecture for
Access Slicing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
5.1.3 Key Techniques for Access Slicing in F-RANs . . . . . . . . . . 88
5.1.4 Challenges and Open Issues. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
5.2 Resource Management in Sliced F-RANs . . . . . . . . . . . . . . . . . . . . . . . . . . 91
5.2.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
5.2.2 System Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
5.2.3 Problem Formulation and Lyapunov Optimization . . . . . . . 95
5.2.4 Solution for Orthogonal and Multiplexed
Subchannel Strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
5.3 Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
Contents xi
6 Dynamic Resource Allocation in Fog Radio Access Networks. . . . . . . . . 105

6.1 Centralized Cost-Aware Energy Efficiency Optimization
in F-RANs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
6.1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
6.1.2 System Model and Problem Formulation . . . . . . . . . . . . . . . . . 106
6.1.3 Problem Solution and Algorithm Design . . . . . . . . . . . . . . . . . . 109
6.2 Cooperative Game Based Interference Management . . . . . . . . . . . . . . . 114
6.2.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
6.2.2 System Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
6.2.3 Coalitional Game Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
6.2.4 Coalition Formation Algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
6.2.5 Simulation Results and Analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . 120
6.3 Deep Reinforcement Learning-Based Resource Management . . . . . 121
6.3.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
6.3.2 System Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
6.3.3 Problem Formulation and Decoupling. . . . . . . . . . . . . . . . . . . . . 124
6.3.4 DRL Based Mode Selection and Resource Management . 125
6.3.5 Simulation Results and Analysis. . . . . . . . . . . . . . . . . . . . . . . . . . . 127
6.4 Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
7 Content Caching in Fog Radio Access Networks . . . . . . . . . . . . . . . . . . . . . . . 133
7.1 Hierarchical Cooperative Content Caching Framework in F-RANs 133
7.2 Content Pushing and Delivering Schemes in F-RANs . . . . . . . . . . . . . 134
7.2.1 System Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
7.2.2 The Conventional Content Pushing and Delivering
Scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
7.2.3 NOMA-Based Content Pushing and Delivering Scheme . 136
7.3 The Performance of Content Pushing and Delivering in F-RANs . 137
7.3.1 The Outage Probability of NOMA-Based Scheme . . . . . . . 137
7.3.2 Further Discussion of Theorem 7.1 . . . . . . . . . . . . . . . . . . . . . . . . 138
7.3.3 Asymptotic Analysis of Outage Probability . . . . . . . . . . . . . . . 141
7.4 Joint Optimization of Cache Management and Resource
Allocation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
7.4.1 Problem Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
7.5 Computation Complexity Analysis and Simulation Results . . . . . . . 149
7.5.1 Computation Complexity Analysis . . . . . . . . . . . . . . . . . . . . . . . . 149
7.6 Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
xii Contents
8 Computation Offloading in Fog Radio Access Networks. . . . . . . . . . . . . . . 153

8.1 Performance Analysis of Hierarchical Cloud-Fog
Computation Offloading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
8.1.1 System Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
8.1.2 Hierarchical Cloud-Fog Computation Offloading
Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
8.1.3 Computation Offloading Probability and Latency
Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
8.1.4 Further Discussion of Offloading Strategy . . . . . . . . . . . . . . . . 161
8.2 Joint Optimization of Computation Offloading in F-RANs . . . . . . . . 162
8.2.1 Problem Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
8.2.2 Resource Management and Offloading Decision
Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
8.3 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
8.3.1 Simulation Results of Latency Performance . . . . . . . . . . . . . . 173
8.3.2 Simulation Results of Joint Optimization Algorithm . . . . . 173
8.4 Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
9 Prototype Design of Fog Radio Access Networks . . . . . . . . . . . . . . . . . . . . . . . 179
9.1 Design Basics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
9.1.1 Enabling Fog Computing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
9.1.2 Network Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
9.2 Useful Development Tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
9.2.1 SDR Software. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
9.2.2 QT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
9.2.3 Database Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
9.2.4 Python . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
9.2.5 TensorFlow. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
9.2.6 Docker. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
9.3 An Example of F-RAN Prototypes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
9.3.1 Hardware Platforms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
9.3.2 Client Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
9.3.3 Featured Functionalities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
9.4 Performance Evaluation of F-RAN Prototypes . . . . . . . . . . . . . . . . . . . . . 197
9.5 Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
Contents xiii
10 Future Trends and Open Issues in Fog Radio Access Networks . . . . . . 203
10.1 Future Trends of F-RANs: Federated Learning-Based Paradigms . 206
10.1.1 The Conventional Federated Learning Paradigm . . . . . . . . . 207
10.1.2 A Hierarchical Federated Learning Paradigm . . . . . . . . . . . . . 208
10.1.3 Potential Applications of Federated Learning in F-RANs 209
10.2 Fundamentals of Federated Learning in F-RANs. . . . . . . . . . . . . . . . . . . 209
10.2.1 Loss Compensation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
10.2.2 Model Compression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
10.2.3 Performance of Loss Compensation and Model
Compression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
10.3 Key Enabling Techniques of Future Evolved F-RANs . . . . . . . . . . . . . 214
10.3.1 Hierarchical Cloud and Fog Computing . . . . . . . . . . . . . . . . . . . 214
10.3.2 Advanced Transmission Techniques . . . . . . . . . . . . . . . . . . . . . . . 214
10.3.3 Resource Management and User Scheduling . . . . . . . . . . . . . 215
10.3.4 Intelligent NFV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
10.4 Open Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
10.4.1 Massive Multiple Access . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
10.4.2 New Theory and Techniques of Deep Learning . . . . . . . . . . . 217
10.4.3 Security and Privacy Issue of Local Model Feedback . . . . 217
10.5 Summary. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
Chapter 1
Brief Introduction of Fog Radio Access
Networks
In Sect. 1.1, we will firstly review the evolution path of radio access networks
(RANs) from heterogeneous networks (HetNets) and cloud radio access networks
(C-RANs) to the fifth generation (5G) and even the sixth generation (6G) mobile
communication systems, and various new emerging services will be introduced,
which show the necessity of presenting fog computing. Then, in Sect. 1.2, the key
features of fog computing and the relationship between fog computing and mobile
edge computing (MEC) will be summarized. Finally, the fog radio access network
(F-RAN) for 5G and 6G will be presented in Sect. 1.3, and the applications as well
as standardization activities with respect to F-RANs will be introduced in Sect. 1.4.
1.1 History and Evolution of RANs
Mobile communication systems based on the cellular technology have a very

tremendous growth in the past 50 years, and the mobile communication generation
generally depends on the changes in terms of system structure, mobility speed, radio
transmit and networking technology, occupied frequency, data capacity, transmit
latency, etc. The first generation (1G) is a basic voice analog phone system based
on frequency division multiple access (FDMA) in the frequency band of 824–
894 MHz with channel bandwidth of 30 KHz. The second generation (2G) uses
a digital technology and supports voice and messaging services based on time
division multiple access (TDMA) and narrow-band code division multiple access
(CDMA). The third generation (3G) allows packet data to transmit the bit rate
up to 14 Mbps operating at the frequency band of 2100 MHz with the frequency
bandwidth of 15–20 MHz, which are mainly used for the high-speed mobile
internet service and multimedia video communication. The fourth generation (4G)
is capable of providing 10 Mbps–1 Gbps peak rate through orthogonal frequency
division multiple access (OFDMA). Based on advanced techniques, such as
© Springer Nature Switzerland AG 2020 1

M. Peng et al., Fog Radio Access Networks (F-RAN), Wireless Networks,
https://doi.org/10.1007/978-3-030-50735-0_1
2 1 Brief Introduction of Fog Radio Access Networks
Fig. 1.1 The evolution of BS structure in cellular systems (Peng et al. 2015b)
multiple-input and multiple-output (MIMO), non-orthogonal multiple access

(NOMA), and millimeter wave communications, 5G opens a new era in mobile
communication technology, which can support ultra-high-speed data (peak capacity
is up to 20 Gbps), ultralow latency (no more than 1 ms), massive connection (at
least 1 million connected devices per square kilometer), and high mobility (up to
500 km/h move speed).
The 2G/3G cellular systems only provide a relatively low-speed (up to 100 kbps
per user) data service over a large coverage area. To increase transmit data rate
for packet service, the wireless local area networks (WLAN) have been presented
to provide a high-speed data service (up to 11 Mbps with IEEE 802.11b standard
and 54 Mbps with IEEE 802.11a standard) over a geographical small area (Peng and
Wang 2009). Inspired by WLAN, the cellular network begins to evolve the functions
of base station (BS), as shown in Fig. 1.1. In 1G/2G cellular systems, radio signal
processing and radio resource management functionalities are integrated together
inside a base station (BS), and all BSs are controlled in a centralized way. However,
in 3G/4G cellular systems, the functions of the traditional BS are divided into
remote radio head (RRH) and baseband unit (BBU). In this new BS architecture,
the location of BBU can be far from RRH for a low site rental and the convenience
of maintenance, and BBUs can be locally coordinated to suppress the inter-cell
interference in a distributed way (Peng et al. 2015b).
The alternative approach to increasing the transmission bit rate is to increase
the transmit power or deployment density of BSs, which creates server inter-cell
interference to other serving user equipment (UEs) in adjacent cells (Hossain 2008).
These traditional cellular systems are reaching their breaking points to provide high
capacity due to the server interference, and the conventional cellular architectures
that are devised to cater to seamless coverage and optimized for high bit rate for
homogeneous traffic have been facing unprecedented challenges and problems.
Different from the seamless coverage, these bursting packet traffics derived from the
mobile internet mainly pursue the high data rate only in some special indoors or hot
spots. As a result, there is an increasing trend to deploy small BSs in hot spots, such
1.1 History and Evolution of RANs 3
as relay stations (RSs), distributed antennas, and access points (APs) for picocells,
femtocells, and small cells. These new small BSs are either operator-deployed or
consumer-deployed, and they have taken big changes to the traditional cellular
architecture, which forms a mix of small cells underlying the macrocells. To address
these challenges and the corresponding changes, heterogeneous network (HetNet)
comprising of marco base stations (MBSs) and small BSs has been presented as an
emerging network paradigm evolution in 4G/5G systems (Vision 2013).
1.1.1 HetNet Overview
HetNet is an advanced networking technology that can cost-efficiently improve

coverage and capacity. The traditional high power nodes (HPNs) are mainly used to
complete the seamless coverage and provide high mobility, while low power nodes
(LPNs), such as small BSs, RSs, APs for picocells, femtocells, and small cells,
are mainly deployed to provide high data rates in some hot spots. By deploying
additional LPNs underlying the HPN, and bringing LPNs close to UEs, HetNets
can potentially improve spatial efficiency through improving the capacity of UEs in
the cell edge, decrease the energy consumption due to short transmit distance, and
keep good connectivity and high mobile capability (Quek et al. 2013).
There are two types of HetNets, i.e., inter-HetNets and intra-HetNets. The inter-
HetNets are presented for the different heterogeneous radio access networks to
interwork and complete cooperative functionalities. For example, the interworking
between WLAN and 3G/4G systems is necessary for users to access the Internet
flexibly. Inter-HetNet can facilitate the flexible utilizations of frequency bands
across different RANs and leverage the frequency spectrum, which have been
widely defined in standards and successfully used in real systems.
Through deploying a large number of LPNs with multi-tier layers in the coverage
holes and hot spots, intra-HetNet is presented to increase the dense reuse of
spectrum and reduce energy consumption for achieving high SE and EE in 4G and
beyond systems. The strategies to reuse the frequency bands among HPNs and LPNs
for intra-HetNets are generally categorized into overlay, underlay, and hybrid. As
shown in Fig. 1.2a, the overlay strategy means that the frequency spectrums are
orthogonally used by HPNs and LPNs, which suggests that the available spectrum
band is not enough, but the cross-tier interference could be avoided under the
low frequency utilization ratio. If users associated with HPNs occupy the spectral
channels, the users associated with LPNs stop to transmit immediately if they also
use the same frequency channels. The overlay strategy is effective in avoiding
the inter-tier interference, but it requires accurate spectrum sensing and complex
cognitive radio.
The underlay strategy shown in Fig. 1.2b is preferred from the operator’s
perspective, where HPNs and LPNs access the whole frequency band with the reuse
way. Both HPNs and LPNs are assigned the same frequency bands under controlling
the severe interference. With the advanced interference coordination techniques,
Fig. 1.2 Three frequency reuse strategies in intra-HetNets
the frequency band can be reused. In particular, the severe inter-tier interference
is hardly coordinated due to the random deployment of LPNs.
As shown in Fig. 1.2c, in the hybrid overlay and underlay strategy, LPNs partially
reuse the spectral resources of HPNs and thus results in the underlay structure.
While the other portion of spectral bands are separately and orthogonally reserved
for HPNs and LPNs, respectively. Considering the balance between performance
gains and implementing complexity, the hybrid strategy is a good solution.
Since the frequency band is scarce, and the underlay HetNet strategy is the most
promising to improve SE and EE, it has already attracted significant attentions
and been defined in 3GPP standards. However, for the successful rollouts, it still
comes with own challenges. As shown in Fig. 1.3, interference coordination and
cancelation (ICC), radio resource allocation optimization (RRAO), cooperative
radio resource management (CRRM), and self-organizing network (SON) are four
key techniques. In terms of ICC, different types of interference should be tackled
in the physical layer, including the inter-tier, inter-cell, and intra-cell interference.
RRAO aims to assign the scarce physical radio blocks (PRBs) to different users
for maximizing SE or EE with low complexity in aspects of the multi-dimensional
resources. Since the resource assignment is strictly related to the radio channel status
and the adopted ICC in physical layer, RRAO often is based on the cross-layer
design mode. In addition, RRAO is responsible for scheduling the PRBs in one
cell, while CRRM mainly tackles the radio resource management among multiple
adjacent cells, which is often relevant to mobility management and soft frequency
reuse. Finally, to reduce the configuration and optimization cost, SON is proposed
to enable AI into the HetNet organization.
Fig. 1.3 System model and key techniques in HetNets
1.1.2 C-RAN Overview
Since most energy is consumed by BSs in the traditional cellular systems, it

is appealing to migrate storage and computation into the “cloud” to create a
“computing entity” so as to optimize the limited radio resource and save the energy
consumption. To tackle the severe interference, C-RAN has been recognized as an
advanced networking architecture, which was mainly proposed by China Mobile
Company. The core idea is to move most signal processing and radio resource
management functions from BSs into BBU pool. The BBU pool can take all
functions of BBU together and curtail both capital and operating expenditures
through cloud computing capabilities while providing high SE and EE (Peng et al.
2016b).
The history of C-RAN developments is shown in Fig. 1.4, and the concept
of C-RAN is firstly proposed with the name of wireless network cloud (WNC)
by IBM Company in 2010 to lower the networking cost and increase networking
flexibility. However, the term C-RAN is formally proposed and fully exploited by
China Mobile Company in 2011. After that, the industry started to research C-
RANs. ZTE Company focused on tackling with the fiber scarcity and presented
several solutions, including the colored fiber connection and optical transport
network bearer (Simeone et al. 2016). To complete the large-scale processing in
the BBU pool, an efficient and scalable GPP was presented by Intel. Followed by
the virtualization techniques, Alcatel-Lucent Company presented the cloud BSs to
decrease consuming the computation resources under guaranteeing the required
Fig. 1.4 Milestones of C-RAN development
Fig. 1.5 System architecture of C-RANs (Peng et al. 2015a)
performances. In 2013, the centralized C-RAN for 4G has been discussed by

NTT DoCoMo Company, and the RAN-as-a-Service (RANaaS) was emphasized
by Telecom Italia Company. In 2014, the centralized C-RAN was discussed in the
white paper of Liquid Radio released by Nokia Networks, and the heterogeneous
C-RAN was firstly proposed by the authors of this book to tackle the challenges of
C-RANs, which can promote C-RANs to evolve 6G.
As shown in Fig. 1.5, the core idea of C-RAN is to decouple functions of
the traditional BSs into several RRHs and a centralized BBU pool. To sup-
port high capacity in hot spots, RRHs with radio frequency functions can be
locally deployed. To suppress the severe inter-tier and inter-cell interferences,
the virtualized BBU pool can take large-scale collaborative processing (LSCP),
cooperative radio resource allocation (CRRA), and intelligent networking through
introducing cloud computing technique. The bottleneck of C-RAN is the fronthaul
that communicates with the BBU pool and RRHs. The fronthaul can be with the
common public radio interface (CPRI) protocol, and it is often capacity and delay
constrained (Peng et al. 2015a).
C-RANs have been advocated by both mobile operators and equipment vendors
due to the potential significant benefits through introducing cloud computing into
cellular networks, but they also come with their own challenges. The biggest
problem is the constraints of capacity and transmit delay from fronthaul links.
Meanwhile, the processing complexity and the corresponding delay in the BBU
pool degrade the performance gains from LSCP and CRRA. Meanwhile, due to the
non-ideal channel status information for all access links among RRHs and active
UEs in the BBU pool, it is hard to make interference mitigate (Park et al. 2013).
The evolution of C-RAN is necessary, but C-RAN’s advantages should be kept in
consideration.
1.1.3 5G Overview
Unlike 2G, 3G, and 4G, 5G is expected to fundamentally transform the role that
telecommunications technology plays in society, which can enable further economic
growth and pervasive digitalization of a hyper-connected society. Not only people
can be connected to 5G via smartphones whenever needed, but also devices,
machines, and even things can create the communicate society through 5G. As
a result, 5G formulates the internet of everything and connects people, devices,
machines, things, data, applications, transport systems, and cities via wireless
networks, which can support a wide variety of applications and services (Chih-Lin
et al. 2014).
There are three typical usage scenarios well defined in 3GPP for 5G, which
includes:
• Enhanced mobile broadband (eMBB): This usage is to deal with hugely increased
data rates, high user density, and high traffic capacity for hot spots scenarios as
well as seamless coverage and high mobility scenarios with still improved used
data rates.
• Massive machine-type communications (MMTC): This usage requires low power
consumption and low data rates for the connected devices. It is mainly with IoT
service.
• Ultra-reliable and low-latency communications (URLLC): This usage caters for
safety-critical and mission-critical applications, such as the massive connections
in the industry IoT.
As shown in Fig. 1.6, 5G is expected to provide 20 times the peak data rate, 10
times lower latency, and 3 times more spectral efficiency than 4G. 5G can transport
a huge amount of data with ultra-high bit rate, reliably connect an extremely
large number of devices, and process high volumes of data with minimal delay.
5G is expected to deliver significantly increased operational performance, i.e.,
increased spectral efficiency, higher data rates, low-latency, as well as superior user
Fig. 1.6 Performance requirements comparisons between 4G and 5G (Chih-Lin et al. 2014)
experience. Meanwhile, 5G is cater not only for eMBB, but also cater for massive
deployment of IoTs, which can offer acceptable levels of energy consumption,
terminal cost, network deployment, configuration, and operation cost.
The increased capacity and peak data rates require more frequency spectrum and
vastly more spectral efficient techniques in 5G than those in 4G. The spectrum bands
allocated for 5G can be divided into three main categories: sub-1 GHz, 1–6 GHz,
and above 6 GHz. Since the propagation properties of the signal benefits to create
large coverage areas and deep in-building penetration, sub-1 GHz bands are often
used to support traditional voice, real-time emerging services, and special services
in high mobility, which can extend the coverage from urban to suburban and rural
areas. The 1–6 GHz bands offer a reasonable mixture of coverage and capacity,
while spectrum bands above 6 GHz provide significant capacity thanks to the very
large bandwidth.
The additional spectrum mainly comes from frequency bands above 24 GHz,
which poses a huge number of challenges from the intrinsic propagation character-
istics of millimeter waves. To enhance transmit performance in the physical layer
of 5G, new radio (NR) has been defined in 3GPP, which is mainly based on flexible
multiple access and coding techniques. Until now, there are two main frequency
bands that have been well defined, i.e., sub-6 GHz and the mmWave range (24–100
GHz) (Series 2015).
Unlike the traditional cellular system that requires both RAN and core network
work in the same generation to be deployed, 5G is expected to integrate elements
of different generations with different configurations. Standalone (SA) mode is
defined for using only one RAN, while non-standalone (NSA) mode is defined
for combining multiple RANs. For NSA, the 5G NR or the evolved 4G LTE radio
cells and the core network can be operated alone, which suggests that the NR or
evolved LTE radio cells can be used for both control and user planes. While SA is
a simple solution for operators, to make user of service continuity, it is deployed
as an independent network through normal inter-generation handover between
4G and 5G.
The initial phase of NSA in 5G mainly focuses on eMBB, which provides high
bit rate complemented by moderate latency improvements and supports several
classical use cases, such as AR/VR, UltraHD, 360-◦ streaming video. MMTC
has been already developed as a part of NB-IoT technologies. A huge number of
MMTC devices connect to the 5G BS, which makes it infeasible to allocate a priori
resources to individual MMTC devices. As a result, the corresponding radio access
mechanisms should be enhanced in 5G. The usage scenario of URLLC supports
low-latency transmissions with small payloads and high reliability, whose bit rate is
relatively low, and the main challenge is to ensure the transmit error rate should be
typically lower than 10−5 .
To satisfy with requirements of the aforementioned three usage scenarios,
especially allow them to coexist with a unified network architecture, network slicing
is urgent, which jointly allocates the resources of communication, cloud computing,
edge computing, fog computing, cloud storage, edge storage, and its aim is to
guarantee the isolation under the required performance levels.
1.1.4 6G Prospect
Since the vision and requirements for 5G defined by International Telecommuni-

cation Union (ITU) were initially issued in 2014, the pace of 5G development was
fast and smooth. In 2016, 5G standardization was formally launched in 3GPP, and
then 5G technology trials were started to conduct by major official organizations
of all over the world. In 2017, the first version of NSA was completely finished,
and then the second version of SA standard was finalized in 2018. In 2019, several
5G pre-commercial networks have been developed widely in the world. Just like
the emergence of smartphones stimulated 3G applications and triggered the demand
for large-scale deployment of 4G, it is believed that some modes of IoT business
will also stimulate the outbreak of 5G industry at some point in the 5G era, thereby
stimulating the demand for the future 6G network (Letaief et al. 2019).
Table 1.1 Possible capabilities of 6G in comparison with 5G

Major Factors 5G 6G
Peak data rate Up to 20 Gbps > 100 Gbps
User experience data rate 1 Gbps > 10 Gbps
Traffic density 10 Tbps/km2 > 100 Tbps/km2
Connection density 1 million/km2 10 million/km2
Delay 10 ms < 1ms
Mobility up to 350 km/h up to 1000 km/h
Spectrum efficiency 3–5x relative to 4G > 3x relative to 5G
Energy efficiency 1000x relative to 4G > 10x relative to 5G
Coverage percent Up to 70% > 99%
Reliability Up to 99.9% > 99.999%
Positioning precision 1m Centimeter level
Receiver sensitivity Up to − 120 dBm < −130 dBm
With the open of the scale-up commercial deployment of 5G, more and more
researchers and related organizations began to consider the evolution of 5G. At
the 2018 Mobile World Congress, an official of the Federal Communications
Commission looked ahead to 6G in public. Not only the USA, China also has
launched 6G related work in March 2018. Consider that wireless communication
systems upgrade to a new generation every 10 years, it can be foreseen that there is
a certain consensus on starting 6G related research since 2020.
The goal of 6G is to meet the needs of the informatization society ten years
later, so the 6G vision should focus on the needs that cannot be satisfied by 5G.
The 6G vision requires massive connectivity, reliability, real-time, and throughput
requirements, which are new and huge challenges to the existing 5G. As shown
in Table 1.1, 6G is expected to upgrade and improve to achieve 10–100 times
higher peak data rate, system capacity, spectrum efficiency, moving speed than 5G.
Meanwhile, it will achieve lower delay, wider and deeper coverage, which enables to
serve the interconnection of everything, fully support the development of intelligent
life and industries. There are several key characteristics that have been general
consent as follows (Chen et al. 2020):
• 6G is expected to be ubiquitous and integrated with broader and deeper coverage
than 5G, including many kinds of communications, such as terrestrial land
communication, space communication, air communication, sea and underwater
communication. With AI driven intent-based networking and network slicing
technologies, 6G can serve in various application scenarios, such as airspace, sky,
land, and sea. In a word, 6G can realize a global ubiquitous mobile broadband
communication system.
• 6G is expected to work on a higher frequency band and a wider bandwidth
to achieve higher peak bit rate and average network capacity than 5G, such as
mmWave communication, TeraHertz communication, visible light communica-
tion, and so on. Compared with 5G, 6G can provide a data rate up to 10–100
times, supporting the peak data rate with 1 Tbps and the user experienced data
rate with 10 Gbps. In addition, 6G can achieve a flexible frequency sharing goal,
which can further enhance the frequency reuse efficiency.
• 6G is expected to be a personalized intelligent and visualized network. Based
on SDN, NFV, SDR, cloud computing, fog computing, and AI techniques,
6G will realize the intent-based, software-defined, flexible, and virtualized net-
working, which depends on the communication, computing, and communication
cooperation. Meanwhile, fog computing will be promising to make cloud and
edge computing adaptive to the application and networking status. As a result,
the traditional centralized 5G will be evolved into the advanced phase of
communication, computing, and communication cooperation. 6G should be data
centralized, content centralized, user centralized, and fully service centralized.
• 6G is expected to have an endogenous security, and the function security is
integrated designed. By introducing blockchain based trust and safety communi-
cation mechanisms, 6G will have the capability of self-awareness, self-analyzed,
self-optimized, self-healing, and self-protect. Both the real-time dynamic analy-
sis and the adaptive risk evaluation will be incorporated, which help realize the
space cybersecurity.
• 6G is expected to merge communication, computation, sensing, and navigation
functions together. To make sure the seamless coverage in sea and mountain,
6G will use satellite communication and make it cooperatively work with
the terrestrial land communication. Meanwhile, the satellite navigation and
positioning systems and even the radar sensing systems will be incorporated.
Based on the open and software-defined networking architecture, 6G can make
networking fast and self-intelligent development.
• 6G is expected to enhance its intelligence via collecting and deeply analyzing
these massive configurations and running data. Meanwhile, 6G can realize
everything intelligence and group collective intelligence, i.e., swarm intelligence.
As shown in Fig. 1.7, from the viewpoint of 3GPP standard organization,
according to the standard scheduling, 3GPP Release 16 mainly for NR techniques
will plan to be finalized in the early 2020, then research on 5G beyond systems
will begin from Release 17 toward Release 19. According to the scheduling, the
key technique research for 6G may be followed from Release 20 about in 2023. In
the ITU standard organization, 5G standards are expected to be formally issued at
the end of 2020, then the corresponding technology research on 6G may be started,
in which the 6G vision and technology trend will be first considered. Meanwhile,
from the viewpoint of industry, the 6G relative research has been started since 2018.
The visions, performance requirements, and key technologies have been discussed
by academics and industries all over the world since 2019. It is expected that these
works will further undergo during 2024–2026, then the standard related works for
6G will be scheduled after 2026, and the final 6G standards will be finalized toward
2030.
The challenges of 6G include system coverage, peak capacity, average user data
rate, transmit delay, movement speed, SE, and EE. 6G will develop the new air
Fig. 1.7 Road map of 6G standards (Letaief et al. 2019)
interface like NR in 5G that enables multiple heterogeneous wireless transmission

accesses. Meanwhile, it will make radio communication to merge computation,
navigation, and sensing, all of which require cloud computing, fog computing, and
AI technology together to empower 6G.
1.2 Fog Computing
Cloud computing is generally used to describe data centers available to massive

users over the Internet. It can be used to save costs and help the customers
focus on their core business instead of deploying massive computing storages
with high IT skills. Thanks to the virtualization technique, cloud computing has
a significant development since NFV and SDN has been presented. Unfortunately,
cloud computing has some drawbacks, especially for the IoT services, including
(Chiang and Zhang 2016):
• High transmit latency: more and more 5G and 6G applications require a low-
latency, but cloud computing cannot guarantee the low transmit latency because
the distance between client devices and data processing centers is relatively long;
• Low computing efficiency: Cloud computing needs an always-on connection
to work properly, which suggests that a failure may reduce the reliability of
the whole network. Meanwhile, interruptions are easy to occur in the cloud
computing-based system and make customers suffer from a large number of
unexpected outages;
• Security and privacy: The private data will be transferred to data processing
centers through globally connected radio and wire channels alongside thousands
of gigabytes of other users’ huge data, and it is vulnerable to cyberattacks and
data loss. Sensitive data should not be transmitted and processed in the cloud
Another random document with
no related content on Scribd:
This threefold classification clearly absorbs the great mass of mankind without straining, but it is
soundest to recognize that this same basic classification requires a certain margin of extensions
along the lines indicated in our table.
The classification made by the French anthropologist Deniker is one of the most elaborate yet
devised. It recognizes 6 grand divisions, 17 minor divisions, and 29 separate races. The primary
criterion of classification is hair texture.
Deniker’s Classification
A. Hair woolly, with broad nose.

I. 1. Bushman.
II. Negroid.
2. Negrito.
3. Negro.
4. Melanesian (including Papuan of New Guinea).
B. Hair curly to wavy.

III. 5. Ethiopian (Sudan, etc.).
IV. 6. Australian.
V. 7. Dravidian (southern India).
VI. 8. Assyroid (Kurds, Armenians, Jews).
C. Hair Wavy.
VII. 9. Indo-Afghan.
VIII. North African.
10. Arab or Semite.
11. Berber (N. Africa).
IX. Melanochroid.
12. Littoral (W. Mediterranean).
13. Ibero-insular (Spain, S. Italy).
14. Western European.
15. Adriatic (N. Italy, Balkans).
D. Hair wavy to straight, with light eyes.

X. Xanthochroid.
16. North European.
17. East European.
E. Hair wavy to straight, with dark eyes.

XI. 18. Ainu.
XII. Oceanian.
19. Polynesian.
20. Indonesian (East Indies).
F. Hair straight.
XIII. American.
21. South American.
22. North American.
23. Central American.
24. Patagonian.
XIV. 25. Eskimo.
XV. 26. Lapp.
XVI. Eurasian.
27. Ugrian (E. Russia).
28. Turco-Tartar (S.W. Siberia).
XVII. 29. Mongol (E. Asia).
In spite of its apparent complexity, this classification coincides quite closely with the
classification which is followed in this book. Inspection reveals that Deniker’s grand division A is
Negroid, C and D Caucasian, F Mongoloid. Of his two remaining grand divisions, B is
intermediate between A and C, that is, between Negroid and Caucasian, and consists of peoples
which are either, like the East Africans, the probable result of a historical mixture of Negroids and
Caucasians, or which, like the Australians, share the traits of both, and are therefore admitted to
have a doubtful status. The other grand division, E, is transitional between Caucasian D and
Mongoloid F, and the peoples of which it consists are those whom we too have recognized as
difficult to assign positively to either stock. In short, Deniker’s classification is much the more
refined, ours the simpler; but essentially they corroborate one another.
31. Other Classifications

Another classification that puts hair texture into the forefront is that of F. Müller. This runs as
follows:
A. Ulotrichi or Woolly-haired.
1. Lophocomi or Tuft-haired: Papua, Hottentot-Bushmen.
2. Eriocomi or Fleecy-haired: African Negroes.
B. Lissotrichi or Straight-haired.
3. Euthycomi or Stiff-haired: Australian, Malay, Mongolian, Arctic, American.
4. Euplocomi or Wavy-haired: Dravidian (S. India), Nubian, (Sudan), “Mediterranean”
(Europe, N. Africa, etc.).
The distinction here made between the Tuft and Fleecy-haired groups is unsound. It rests on a
false observation: that a few races, like the Bushmen, had their head-hair growing out of the scalp
only in spots or tufts. With the elimination of this group, its members would fall into the Fleecy or
Woolly-haired one, which would thus comprise all admitted Negroids; whereas the two remaining
groups, the Stiff and Wavy-haired, obviously correspond to the Mongoloid and Caucasian. The
only remaining peculiarity of the classification—and in this point also it is unquestionably wrong—
is the inclusion of the Australians in the Stiff or Straight-haired group. But even this error reflects
an element of truth: it emphasizes the fact that in spite of their black skins, broad noses, and
protruding jaws, the Australians are not straight-out Negroids.
The underlying feature of this classification, after allowing for its errors, is that mankind consists
of two rather than three main branches: the Ulotrichi or Negroids, as opposed to the Lissotrichi or
combined Mongoloids and Caucasians. This basic idea has been advocated by others. Boas, for
instance, reckons Mongoloids and Caucasians as at bottom only subtypes of a single stock with
which the Negroids and Australians are to be contrasted.
Somewhat different in plan is Huxley’s scheme, which recognizes four main races, or five
including a transitional one. These are (1) Australioids, including Dravidians and Egyptians; (2)
Negroids, with the Bushmen and the Oceanic Papuans, Melanesians, Tasmanians, and Negritos
as two subvarieties; (3) Mongoloids, as customarily accepted; (4) Xanthochroi, about equivalent
to Nordics and Alpines; (5) Melanochroi, nearly the same as the Mediterraneans, but supposed by
Huxley to be hybrid or intermediate between the Xanthochroi and Australioids. This classification
in effect emphasizes the connection between Australoids and Caucasians, with the Negroids as a
distinctive group on one side and the Mongoloids on the other.
Haeckel’s classification is basically similar, in that besides the usual three primary stocks—
which he elevates into species—he recognizes a separate group comprising the Australians,
Dravidians, and Vedda-like Indo-Australians.
32. Principles and Conclusions Common to All Classifications

It will be seen that in spite of the differences and uncertainties as yet prevalent in any scheme
for classifying the human species, certain principles stand out both as regards method and
results; and in regard to these principles there is substantial agreement.
First, any valid classification must rest on a combination of as many traits or features as
possible.
Second, several features of the human body are of definite significance for the discrimination of
races. Hair and hairiness are unquestionably of great importance; stature, except in extreme
cases, much less so. Color differences in the skin, hair, or eyes are important but difficult to
handle. Shape of nose and prognathism are useful for rough classification. The cephalic index
possesses an exceptional utility in making the finer discriminations.
Third, it is clearly impossible to find a simple and consistent scheme within which all the
varieties of man can be placed. We must not attempt more than nature allows.
On the other hand the vast bulk of mankind does fall naturally into three great divisions, each of
which again subdivides into three or four principal branches, in regard to whose distinctness there
is no serious difference of opinion. The scattering remainder of races are allied sometimes to one
primary stock, sometimes to another, but always with some special peculiarities.
From such a classification as this, especially after the accumulation of large series of accurate
measurements which will permit its being worked out to greater exactness, we may hope
ultimately to reconstruct the full and true history of the races of men, or, in any event, some
reasonable hypothesis as to their development. As yet, however, we are not in a position to
account for the origin of the races except speculatively.
33. Race, Nationality, and Language

The term race has here been used in its biological sense, for a group united in blood or
heredity. A race is a subdivision of a species and corresponds to a breed in domestic animals.
Popularly, the word is used in a different sense, namely that of a population having any traits in
common, be they hereditary or non-hereditary, biological or social (Chapter I). It is customary, but
scientifically inaccurate, to speak of the French race, the Anglo-Saxon race, the Gypsy race, the
Jewish race. The French are a nation and nationality, with a substantially common speech;
biologically, they are three races considerably mixed, but still imperfectly blended (§ 24). Anglo-
Saxon refers primarily to speech, incidentally to a set of customs, traditions, and points of view
that are more or less associated with the language. The Gypsies are a self-constituted caste, with
folkways, occupations, and a speech of their own. The Jews, who were once a nationality, at
present, of course, form a religious body, which somewhat variably, in part from inner cohesion
and in part from outer pressure, tends also to constitute a caste. They evince little hereditary
racial type, measurements indicating that in each country they approximate the physical type of
the gentile population.
It may seem of little moment whether the word race is restricted to its strict biological sense or
used more loosely. In fact, however, untold loose reasoning has resulted from the loose
terminology. When one has spoken a dozen times of “the French race,” one tends inevitably to
think of the inhabitants of France as a biological unit, which they are not. The basis of the error is
confusion of organic traits and processes with superorganic or cultural ones; of heredity with
tradition or imitation. That civilizations, languages, and nationalities go on for generations is
obviously a different thing from their being caused by generation. Slovenly thought, tending to
deal with results rather than causes or processes, does not trouble to make this discrimination,
and every-day speech, dating from a pre-scientific period, is ambiguous about it. We say not only
“generation,” when there is no intent to imply the reproductive process, but “good breeding”
(literally, good brooding or hatching or birth), when we mean good home training or education; just
as we “inherit” a fortune or a name—social things—as well as ineradicable traits like brown eye-
color. Biology has secured for its processes the exclusive use of the term “heredity”; and
biologists employ the term “race” only with reference to a hereditary subdivision of a species. It is
equally important that the word be used with the same exact denotation in anthropology, else all
discussion of race degenerates irretrievably into illogical sliding in and out between organic and
social factors. The inherently great difficulties which beset the understanding and solution of what
are generally called race problems, as discussed in the next chapter, are considerably increased
by a confusion between what is and what is not racial and organic and hereditary.
CHAPTER IV
PROBLEMS OF RACE
34. Questions of endowment and their validity.—35. Plan of inquiry.—36.

Anatomical evidence on evolutionary rank.—37. Comparative
physiological data.—38. Disease.—39. Causes of cancer incidence.
—40. Mental achievement and social environment.—41.
Psychological tests on the sense faculties.—42. Intelligence tests.—
43. Status of hybrids.—44. Evidence from the cultural record of races.
—45. Emotional bias.—46. Summary.
34. Questions of Endowment and Their

Validity
Are the human races alike or dissimilar in mentality and character?
Are some lower than others, or are they all on a plane as regards
potentiality? The answers to these questions are of theoretical
import, and naturally also bear on the solution of the practical race
problems with which many nations are confronted.
As long as an inquiry remains sufficiently abstract or remote, the
desirability of such inquiry is likely to go unquestioned. As soon,
however, as investigation touches conduct—for instance, our actual
relations with other races—a sentiment has a way of rising, to the
effect that perhaps after all the problem does not so much call for
knowledge as for action. Thus, in regard to the negro problem in the
United States, it is likely to be said that the immediate issue is what
may be the best attitude toward “Jim Crow” cars and other forms of
segregation. Are these desirable or undesirable, fair or unfair? Here
are specific problems which an actual condition presses to have
answered. Under the circumstances, it will be said, is not an inquiry
into the innate capacity of the negro rather remote, especially when
every one can see by a thousand examples that the negro is
obviously inferior to the Caucasian? He is poorer, more shiftless,
less successful. He has made no inventions, produced no geniuses.
He clearly feels himself inferior and comports himself accordingly.
Why then raise the issue of capacity at all, unless from a desire to
befog it, to subvert the conclusions of common sense and every-day
experience by special pleading which substitutes adroitness for
sincerity? When a prisoner has been found guilty it is the judge’s
business to determine the length of sentence, to decide how far
justice should be tempered with mercy. Were he to reopen the case
from the beginning, he would be showing partiality. Is not the
situation of the scientist proposing to inquire into the accepted
verdict that the negro is inferior to the Caucasian, analogous to that
of a judge who insists on setting aside the verdict of twelve
unprejudiced jurymen in order to retry the defendant himself? In
some such form as this, objections may rise in the minds of some.
The answer to such criticism is first of all that racial inferiority and
superiority are by no means self-evident truths. Secondly, the belief
in race inequalities is founded in emotion and action and then
justified by reasoning. That is, the belief is rationalized, not primarily
inferred by pure reason. It may be true, but it is not proved true.
As to what is self-evident, there is nothing so misleading as direct
observation. We see the sun move and the earth stand still. It is
“self-evident” that the sun revolves around the earth. Yet after
thousands of years the civilized portion of mankind finally came to
believe that it was the earth that spun. Science had no perverse
interest, no insidious motive, in advocating the Copernican instead of
the Ptolemaic system; in fact, was driven to its new belief gradually
and reluctantly. It was pre-scientific humanity, with its direct,
homespun, every-day observation, which had really prejudged the
matter, and which, because it had always assumed that the earth
was flat and stationary, and because every idiot could see that it was
so, long combated the idea that it could be otherwise.
As to opinions founded in emotion and subsequently rationalized,
instead of being evolved by pure reason from evidence, it may
suffice to quote from a famous book on herd instinct, as to the
relation of mass opinion and science:
“When, therefore, we find ourselves entertaining an opinion about
the basis of which there is a quality of feeling which tells us that to
inquire into it would be absurd, obviously unnecessary, unprofitable,
undesirable, bad form, or wicked, we may know that that opinion is a
non-rational one, and probably, therefore, founded upon inadequate
evidence.
“Opinions, on the other hand, which are acquired as the result of
experience alone do not possess this quality of primary certitude.
They are true in the sense of being verifiable, but they are
unaccompanied by that profound feeling of truth which belief
possesses, and, therefore, we have no sense of reluctance in
admitting inquiry into them. That heavy bodies tend to fall to the
earth and that fire burns fingers are truths verifiable and verified
every day, but we do not hold them with impassioned certitude, and
we do not resent or resist inquiry into their basis; whereas in such a
question as that of the survival of death by human personality we
hold the favorable or the adverse view with a quality of feeling
entirely different, and of such a kind that inquiry into the matter is
looked upon as disreputable by orthodox science and as wicked by
orthodox religion. In relation to this subject, it may be remarked, we
often see it very interestingly shown that the holders of two
diametrically opposed opinions, one of which is certainly right, may
both show by their attitude that the belief is held instinctively and
non-rationally, as, for example, when an atheist and a Christian unite
in repudiating inquiry into the existence of the soul.”
Take the attitude of the average Californian or Australian about the
Mongolian; of the Texan about the Mexican; of the Southerner about
the Negro; of the Westerner about the local tribes of Indians; of the
Englishman about the Hindu—is not their feeling exactly described
by the statement that inquiry into the possibility of racial equality
would be “unnecessary,” “absurd,” or evilly motivated; and that their
belief in race superiority rests on an “a priori synthesis of the most
perfect sort,” and possesses “the quality of primary certitude”?
In short, the apparently theoretical beliefs held as to race capacity
by people who are actually confronted by a race conflict or problem
are by no means the outcome of impartial examination and
verification, but are the result of the decisions taken and emotions
experienced in the course of acts performed toward the other race.
The beliefs rest ultimately on impulse and feeling; their reasoned
support is a subsequent bolstering up. Of course, the fact that a
belief springs from emotion does not render that belief untrue, but
does leave it scientifically unproved, and calling for investigation.
These conclusions may vindicate inquiry into the relative capacity
of races from the charge of being finespun, insidious, impractical, or
immoral.
35. Plan of Inquiry

In approach to the problem, a consideration stands out. If the
human races are identical in capacity, or if, though not absolutely
alike, they average substantially the same in the sum total of their
capacities, then such differences as they have shown in their history
or show in their present condition must evidently be the result mainly
of circumstances external to heredity. In that case, knowledge of the
historical or environmental circumstances, and analysis of the latter,
become all-important to understanding. On the other hand, if
hereditary racial inequalities exist, one can expect that the historical
or cultural influences, however great they may be, will nevertheless
tend to have their origin in the hereditary factors and to reinforce
them. In that case, differences between two groups would be due
partly to underlying heredity and partly to overlying cultural forces
tending on the whole in the same direction. Yet even in that case,
before one could begin to estimate the strength of the true racial
factors, the historical ones would have to be subtracted. Thus, in
either event, the first crux of the problem lies in the recognition and
stripping off of cultural, social, or environmental factors, so far as
possible, from the complex mass of phenomena which living human
groups present. In proportion as these social or acquired traits can
be determined and discounted, the innate and truly racial ones will
be isolated, and can then be examined, weighed, and compared.
Such, at any rate, is a reasonable plan of procedure. We are looking
for the inherent, ineradicable elements in a social animal that has
everywhere built up around himself an environment—namely, his
culture—in which he mentally lives and breathes. It is precisely
because in the present inquiry we wish to get below the effects of
culture that we must be ready to concern ourselves considerably
with these effects, actual or possible.
36. Anatomical Evidence on Evolutionary

Rank
But first of all it may be well to consider the relatively simple
evidence which has to do with the physical form and structure of
race types. If one human race should prove definitely nearer to the
apes in its anatomy than the other races, there would be reason to
believe that it had lagged in evolution. Also there would be some
presumption that its arrears were mental as well as physical.
But the facts do not run consistently. One thinks of the Negro as
simian. His jaws are prognathous; his forehead recedes; his nose is
both broad and low. Further, it is among Caucasians that the
antithetical traits occur. In straightness of jaws and forehead,
prominence and narrowness of nose, Caucasians in general exceed
the Mongoloids. Thus the order as regards these particular traits is:
ape, Negroid, Mongoloid, Caucasian. With ourselves at one end and
the monkey at the other, the scale somehow seems right. It appeals,
and seems significant. Facts of this sort are therefore readily
observed, come to be remembered, and rise spontaneously to mind
in an argument on race differences.
However, there are numerous items that conflict with this
sequence. For instance, one of the most conspicuous differences of
man from the apes is his relative hairlessness. Of the three main
stocks, however, it is the Caucasian that is the most hairy. Both
Mongoloids and Negroids are more smooth-skinned on face and on
body.
In hair texture, the straight-haired Mongoloid is nearest the apes,
the wavy-haired Caucasian comes next, and the woolly Negroid is
the most characteristically human, or at least unsimian.
In the length of head hair, in which man differs notably from the
monkeys, the relatively short-haired Negro once more approximates
most closely to the ape, but the long-haired Mongoloid surpasses the
intermediate Caucasian in degree of departure.
Lip color reverses this order. The apes’ lips are thin and grayish;
Mongoloid lips come next; then those of Caucasians; the full, vivid,
red lips of the Negro are the most unapelike of all.
It is unnecessary to multiply examples. If one human racial stock
falls below others in certain traits, it rises above them in other
features, insofar as “below” and “above” may be measurable in
terms of degree of resemblance to the apes. The only way in which a
decision could be arrived at along this line of consideration would be
to count all features to see whether the Negro or the Caucasian or
the Mongoloid was the most unapelike in the plurality of cases. It is
possible that in such a reckoning the Caucasian would emerge with
a lead. But it is even more clear that whichever way the majority fell,
it would be a well divided count. If the Negro were more apelike than
the Caucasian in all of his features, or in eight out of ten, the fact
would be heavily significant. With his simian resemblances
aggregating to those of the Caucasian in a ratio of say four to three,
the margin would be so close as to lose nearly all its meaning. It is
apparently some such ratio as this, or an even more balanced one,
that would emerge, so far as we can judge, if it were feasible to take
a census of all features.
It should be added that such a method of comparison as this
suffers from two drawbacks. First, the most closely related forms
now and then diverge sharply in certain particulars; and second, a
form which on the whole is highly specialized may yet have
remained more primitive, or have reverted to greater primitiveness in
a few of its traits, than relatively unevolved races or species.
Thus, the anthropoid apes are brachycephalic, but all known types
of Palæolithic man are dolichocephalic. Matched against the apes,
the long-headed Negro would therefore seem to be the most
humanly specialized stock. Compared however with the fossil human
forms, the Negro is the most primitive in this feature, and the
Mongoloid and Alpine Caucasian could be said to have evolved the
farthest because their heads are the roundest. Yet their degree of
brachycephaly is approximately that of the anthropoid apes. To
which criterion shall be given precedence? It is impossible to say.
Quite likely the round-headedness of the apes represents a special
trait which they acquired since their divergence from the common
hominid ancestral stem. If so, their round-headedness and that of the
Mongoloids is simply a case of convergent evolution, of a character
repeating independently, and therefore no evidence of Mongoloid
primitiveness. Yet, if so, the long-headedness common to the early
human races and the modern Negroids would probably also mean
nothing.
It is even clearer that other traits have been acquired
independently, have been secondarily evolved over again. Thus the
supraorbital ridges. When one observes the consistency with which
these are heavy in practically all Neandertal specimens; how they
are still more conspicuous in Pithecanthropus and Rhodesian man;
how the male gorilla shows them enormously developed; and that
among living races they are perhaps strongest in the lowly
Australian, it is tempting to look upon this bony development as a
definite sign of primitiveness. Yet there is an array of contradicting
facts. The youthful gorilla and adult orang are without supraorbital
development. The male gorilla has his powerful brows for the same
reason that he has the crest along the top of his skull: they are
needed as attachments for his powerful musculature. They are
evidently a secondary sex character developed within the species.
So among fossil men there seem to have been two strains: one
represented by Pithecanthropus and Neandertal man and the
Rhodesian race, which tended toward supraorbital massiveness; and
another, of which Piltdown man is representative, which was smooth
of forehead. Among living races the Asiatic Mongoloids lack marked
supraorbital development; the closely related American Indians
possess it rather strongly; Caucasians and Negroes show little of the
feature; Australians most of all. Evidently it would be unsafe to build
much conclusion on either the presence or absence of supraorbital
ridges.
Perhaps these instances will suffice to show that even the mere
physical rating of human races is far from a simple or easy task. It is
doubtful whether as yet it is valid to speak of one race as physically
higher or more advanced, or more human and less brutish, than
another. This is not an outright denial of the possibility of such
differential ratings: it is a denial only of the belief that such
differentials have been established as demonstrable.
37. Comparative Physiological Data

There is another angle of approach. This consists in abandoning
the direct attempt to rate the races in anatomical terms, and inquiring
instead whether they show any physiological differences. If such
differences can be found, they may then perhaps be interpretable as
differences in activity, responsiveness, endurance, or similar
constitutional qualities. If the bodies of two races behave differently,
we should have considerable reason to believe that their minds also
behaved differently.
Unfortunately, we possess fewer data on comparative physiology
than on comparative anatomy. The evidence is more fluctuating and
intricate, and requires more patience to assemble. Unfortunately,
too, for the purposes of our inquiry, the races come out almost
exactly alike in the simpler physiological reactions. The normal body
temperature for Caucasian adults is 37° (98.5 F.), the pulse about
70, the respiration rate around 17 or 18 per minute. If the Negro’s
temperature averaged even a degree higher, one might expect him
to behave, normally, a little more feverishly, to respond to stimulus
with more vehemence, to move more quickly or more restlessly. Or,
if the pulse rate of Mongolians were definitely lower, they might be
expected to react more sluggishly, more sedately, like aging
Caucasians. But such observations as are available, though they are
far from as numerous as is desirable, reveal no such differences:
temperature, pulse, respiration, record the same as among
Caucasians, or differ so slightly, or so conflictingly, as to leave no
room for positive conclusions. Certainly if there existed any important
racial peculiarities, they would have been noted by the physicians
who at one time or another have examined millions of Negroes,
Chinese, Japanese, and thousands of Indians and Polynesians.
Apparently there is only one record that even hints at anything
significant. Hrdlička, among some 700 Indians of the Southwestern
United States and Northwestern Mexico, found the pulse to average
about 60 per minute, or ten beats less than among whites. This
would seem to accord with the general impression of Indian
mentality as stolid, reserved, slow, and steady. But the number of
observations is after all rather small; the part of the race represented
by them is limited; and the habitat of the group of tribes is mostly a
high plateau, and altitude notoriously affects heart action.
Considerable corroboration will therefore be needed before any
serious conclusions can be built upon this suggestive set of data.
There are other physiological functions that are likely to mean
more than the rather gross ones just considered: for instance, the
activity of the endocrines or glands of internal secretion. An excess
or deficiency of activity of the thyroid, pituitary, adrenals, and sex
glands affects not only health, but the type of personality and its
emotional and intellectual reactions. For example, cretinism with its
accompaniment of near-idiocy is the result of thyroidal under-
development or under-functioning, and is often cured by supplying
the lack of thyroidal substance and secretion. But this subject is as
difficult as it is interesting; to date, absolutely nothing is known about
endocrine race differences. It would be a relatively simple matter to
secure first-hand information on the anatomy of the endocrine glands
in Negroes as compared with whites; to ascertain whether these
differed normally in size, weight, shape, or structure, and how. But
this knowledge has scarcely been attempted systematically, and still
less is any knowledge available in the more delicate and complex
field of the workings of the organs. To be sure, theories have been
advanced that race differentiation itself may be mainly the result of
endocrine differentiations. There is something fascinating about such
conjectures, but it is well to remember that they are unmitigated
guesses.
38. Disease
Pathology might seem to promise more than normal physiology.
So far as mortality goes, there are enormous differences between
races. And the mortality is often largely the result of particular
diseases. Measles, for instance, has often been a deadly epidemic
to uncivilized peoples, and smallpox has in some regions at times
taken toll of a quarter of the population in a year or two. Yet it is
short-sighted to infer from such cases any racial predisposition or
lack of resistance. The peoples in question have been free for
generations, perhaps for their entire history, from these diseases,
and have therefore not maintained or acquired immunity. Their
difference from us is thus essentially in experience, not hereditary or
racial. This is confirmed by the fact that after a generation or two the
same epidemics that at first were so deadly to Polynesians or
American Indians sink to almost the same level of mild virulence as
they show among ourselves.
Then, too, immediate environment plays a part. The savage often
has no idea of contagion, and still less of guarding against it; he
thinks in terms of magic instead of physiology—and succumbs. How
far heavy mortality is the result of lack of resistance or of
fundamentally vicious treatment, is often hard to say. If we tried to
cure smallpox by subjecting patients to a steam-bath and then
having them plunge into a wintry river, we should perhaps look upon
the disease as a very nearly fatal one to the Caucasian race.
39. Causes of Cancer Incidence

It may be worth while to consider briefly the facts as to mortality
from cancer. This dread disease appears to be not contagious, so
that the factor of acquired immunity is eliminated. It is regarded as
incurable, except by operation, so that differences in treatment
become relatively unimportant. If therefore significant differences in
racial liability to cancer exist, they should emerge with unusual
clearness and certainty.
At first sight they seem to. It has been alleged that the white race
is the most susceptible to this affliction. The supporting figures are
as follows: cancer deaths per year per 100,000 population.
1906-10 Denmark 137
England 94
United States 73
1909-11 Johannesburg, whites 52
Negroes 14
1906-10 Natal, Europeans 56
East Indians 11
1906-10 Hongkong, Europeans 53
Chinese 5
1912 Dutch East Indies, Europeans 81
1906-10 Singapore, natives 13
Straits Settlements, natives 10
Ceylon, natives 5
Calcutta, natives 11
1908-13 Manila, whites 51
Filipinos 27
Chinese 19
1910-12 United States, whites 77
Negroes 56
1914 United States, Indians 4
It would seem from these figures that Caucasians die more

frequently of cancer than members of the darker races. In fact, this
has been asserted. Let us however continue with figures.
1908-12 Large cities, latitudes 60°-50° North 106

50°-40° ” 92
40°-30° ” 78
30° North-30° South 38-42
30°-40° South 90
This table would make cancer mortality largely a function of

geographical latitude, instead of race.
Another factor enters: occupation. The following data give the
death rate per 100,000 population among males of 45-54 in England
and Wales.
1890-92 1900-02
Lawyers 199 159
Physicians 102 121
Clergymen 81 91
Chimneysweeps 532 287
Brewers 190 239
Metal workers 120 137
Gardeners 88 93
All occupations 118 145
That the relative incidence is more than a temporary accident is

shown by the approximate recurrence of the frequencies after ten
years.
In proportion as latitude and occupation influence the occurrence
of cancer, race is diminished as a cause. It is reduced still further by
other considerations. The rate for Austria in 1906-10 was 78, for
Hungary 44. Here the race is the same: the difference must be
social. Austria averaged higher in wealth, education, medical
development. This fact would tend to have a double effect. First,
among the more backward population, a certain proportion would die
of internal cancers difficult to diagnose, without the cause being
recognized, owing to insufficient medical treatment. Second, the
general death rate would be higher. More children and young people
would die of infectious or preventable disease, leaving fewer
survivors to die of cancer in middle and old age. Wherever, on the
other hand, a public is medically educated, and typhoid, smallpox,
diphtheria, tuberculosis claim fewer victims, the proportion of those
dying of cancer, nephritis, heart diseases, increases. Such an
increase is noted everywhere, and goes hand in hand with a longer
average life. The alarm sometimes felt at the modern “increase” of
cancer is therefore unfounded, because it is perhaps mainly
apparent. If a larger percentage of the population each year died of
old age, it would be a sign that sanitation and medicine were
increasingly effective: evidence that more people lived to become
old, not that age debility was spreading.
Consequently, a high degree of modern civilization must tend to
raise the cancer rate; and any group of people will seem relatively
immune from cancer in proportion as they remain removed from
attaining to this civilization. In Hungary, from 1901-04, the cancer
deaths were 239 among the owners of large farms, 41 among the
owners of small farms; 108 among employing blacksmiths, 25
among their employees; 114 among employing tailors, 32 among
employed tailors. Obviously these pairs of groups differ chiefly in
their economic and cultural status.
Here too lies the explanation of why the South African negro
shows a rate of only 14, the United States negro of 56; also why the
Chinese rate is as low as 5 in Hongkong, rises to 19 in Manila, and
26 in Hawaii, while the closely allied Japanese average 62 for the
whole of Japan—as compared with 50 for Spain, which is pure
Caucasian, but one of the most backward countries in Europe. In
Tokyo and Kyoto the rate soars to 73 and 90 respectively, just as in
the United States it is about 10 higher for the urban than for the rural
population.
Within the United States, also, the rate rises and falls almost
parallel for whites and Negroes according to locality; as,
1906-10 White Negro

Memphis 59 34
Charleston 73 37
Nashville 74 55
New Orleans 86 73
If allowance is made for the facts that the negro population of the
United States is poorer and less educated than the white; that it lives
mainly in lower latitudes; and that it tends to be rural rather than
urban, the comparative cancer death rates for the country of negro
56 and white 77 would appear to be accounted for, without bringing
race into consideration.
In short, what at first glance, or to a partisan pleader, would seem
to be a notable race difference in cancer liability, turns out so
overwhelmingly due to environmental and social causes as to leave
it doubtful whether racial heredity enters as a factor at all. This is not
an assertion that race has nothing whatever to do with the disease; it
is an assertion that in the present state of knowledge an inherent or
permanent connection between race and cancer incidence has not
been demonstrated. If there is such a connection, it is evidently a
slight one, heavily overlaid by non-racial influences; and it may be
wholly lacking.
The case would be still less certain for most other diseases, in
which environmental factors are more directly and obviously
influential. Racial medical science is not impossible; in fact it should
have an important future as a study; but its foundations are not yet
laid.
40. Mental Achievement and Social

Environment
One point will have become clear in the course of the foregoing
discussion: namely, how far the difficulty of coming to positive
conclusions is due to the two sets of interacting causal factors, the
hereditary ones and the environmental ones that play upon heredity.
The environmental factors are themselves a composite of
geographical influences and of the economic, cultural, and other
social influences that human beings exert upon each other.
If this intermingling of distinct kinds of causes is true of races when
considered from the side of physiology and medicine, it is evident
that the intermingling will be even more intricate in the mental
sphere. After all, bodily functioning varies only within fairly definite
limits. When external influences press too strongly upon the innate
nature of the organism, the latter ceases to function and dies. The
mind, on the other hand, however much its structure may be given
by heredity, depends for its content wholly on experience, and this
experience can be thoroughly varied. Individuals of the same organic
endowment may conceivably be born either in the uppermost
stratum of a highly refined civilization, or among the most backward
and remote savages. Whether this actually happens, and to what
degree, is of course precisely the problem which we are trying to
solve. But that it is theoretically and logically possible cannot be
denied; and here a vicious circle of reasoning begins. One argument
says: there have been no recognized geniuses among peoples like
the Hottentots, and the sum total of their group achievement is
ridiculously small; therefore it is clear that the Hottentot mind must
be inferior. The opposite argument runs: Hottentot cultural
environment is so poor and limited that the finest mind in the world
reared under its influence would grow up relatively sterile and
atrophied; therefore it is probable that the mind of the Hottentot is
intrinsically identical with our own, or at least of equivalent capacity,
and that Hottentot geniuses have actually been born but have been
unable to flourish as geniuses.
Evidently the same facts are before those who advocate these
opposite views, but these facts are viewed from diametrically
opposite sides. If one starts to travel around the logical circle in one
direction, one can keep revolving indefinitely and find ever fresh
supporting evidence. If, however, one begins to revolve around the
same circle of opinion in the opposite direction, it is just as easy and
just as compelling to continue to think in this fashion and to find all
testimony corroborative.
In such a situation it is possible to realize that from the point of
view of proof, or objective truth, one view is worth as much as the
other: which is nothing. It is an emotional bias that inclines one man
toward the conviction of race superiority and another to that of race
equality. The proofs in either case are for the most part a mere
assembling of ex parte testimony. It is easy enough to advocate
impartiality. The difficulty is in being impartial; because both the
hereditary and the environmental factors are in reality unknown
quantities. What we have objectively before us is such and such a
race or group of people, with such and such present traits and
historical record. These phenomena being the product of the
interaction of the two sets of causes, we could of course, if we knew
the strength of one, compute the strength of the other. But as we
have isolated neither, we are dealing with two indeterminate
variables. Evidently the only way out of the dilemma, at any rate the
only scientific way, is to find situations in which one of the factors is,
for the time being, fixed. In that case the strength of the other factor
will of course be proportionate to the attainments of the groups.
Actually, such instances are excessively difficult to find. There are
occasional individuals with identical heredity, namely, twins produced
from the division of a single ovum. In such twins, the strength of
environmental influences can be gauged by the difference in their
careers and achievements. Yet such twins are only individuals, and it
is illegitimate to make far-reaching inferences from them to larger
groups, such as the races. It is conceivable that heredity might on
the whole be a more powerful cause than environment, and racial
groups still average substantially alike in their heredity. Because a
natively gifted and a natively stunted individual within the group vary
conspicuously in achievement, even under similar environment, it
does not follow that races differ in germ-plasm because they differ in
achievements.
If, on the other hand, one sets out to discover cases of identical
environment for distinct racial strains, the task quickly becomes even
more difficult. Very little analysis usually suffices to show that the
environment is identical only up to a certain point, and that beyond
this point important social divergences begin. Thus, so far as
geographical environment goes, the Negro and the white in the
southern United States are under the same conditions. There is also
uniformity of some of the gross externals of cultural environment.
Both Negroes and whites speak English; are Christians; plant corn;
go to the circus; and so on. But, just as obviously, there are aspects
in which their social environment differs profoundly. Educational
opportunities are widely different. The opportunity of attaining
leadership or otherwise satisfying ambition is wide open to the white,
and practically closed to the Negro. The “color-line” inevitably cuts
across the social environment and makes of it two different
environments.

PDF Fog Radio Access Networks F RAN Architectures Technologies and Applications Mugen Peng Download

Uploaded by

Copyright:

Available Formats

PDF Fog Radio Access Networks F RAN Architectures Technologies and Applications Mugen Peng Download

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

PDF Fog Radio Access Networks F RAN Architectures Technologies and Applications Mugen Peng Download

Uploaded by

Copyright:

Available Formats

Download and Read online, DOWNLOAD EBOOK, [PDF EBOOK EPUB ], Ebooks

download, Read Ebook EPUB/KINDE, Download Book Format PDF

Fog Radio Access Networks F RAN Architectures

5G radio access networks : centralized RAN, cloud-RAN,

Blockchain-enabled Fog and Edge Computing: Concepts,

Blockchain-enabled Fog and Edge Computing: Concepts,

Spectrum Access and Management for Cognitive Radio

Elastic Optical Networks Architectures Technologies and

Privacy Enhancing Fog Computing and Its Applications

Media Convergence and the Development Strategies of

Media Convergence and the Development Strategies of

More information about this series at http://www.springer.com/series/14180

Fog Radio Access Networks

ISSN 2366-1186 ISSN 2366-1445 (electronic)

© Springer Nature Switzerland AG 2020

Beijing, China Mugen Peng

1 Brief Introduction of Fog Radio Access Networks . . . . . . . . . . . . . . . . . . . . . . 1

3 Theoretical Performance Analysis of Fog Radio Access Networks . . . 41

6 Dynamic Resource Allocation in Fog Radio Access Networks. . . . . . . . . 105

8 Computation Offloading in Fog Radio Access Networks. . . . . . . . . . . . . . . 153

1.1 History and Evolution of RANs

Mobile communication systems based on the cellular technology have a very

© Springer Nature Switzerland AG 2020 1

multiple-input and multiple-output (MIMO), non-orthogonal multiple access

1.1.1 HetNet Overview

HetNet is an advanced networking technology that can cost-efficiently improve

Fig. 1.2 Three frequency reuse strategies in intra-HetNets

Fig. 1.3 System model and key techniques in HetNets

1.1.2 C-RAN Overview

Since most energy is consumed by BSs in the traditional cellular systems, it

Fig. 1.4 Milestones of C-RAN development

Fig. 1.5 System architecture of C-RANs (Peng et al. 2015a)

performances. In 2013, the centralized C-RAN for 4G has been discussed by

Since the vision and requirements for 5G defined by International Telecommuni-

Table 1.1 Possible capabilities of 6G in comparison with 5G

Fig. 1.7 Road map of 6G standards (Letaief et al. 2019)

interface like NR in 5G that enables multiple heterogeneous wireless transmission

1.2 Fog Computing

Cloud computing is generally used to describe data centers available to massive

A. Hair woolly, with broad nose.

B. Hair curly to wavy.

D. Hair wavy to straight, with light eyes.

E. Hair wavy to straight, with dark eyes.

31. Other Classifications

32. Principles and Conclusions Common to All Classifications

33. Race, Nationality, and Language

34. Questions of endowment and their validity.—35. Plan of inquiry.—36.

34. Questions of Endowment and Their

35. Plan of Inquiry

36. Anatomical Evidence on Evolutionary

37. Comparative Physiological Data

39. Causes of Cancer Incidence

It would seem from these figures that Caucasians die more

1908-12 Large cities, latitudes 60°-50° North 106

This table would make cancer mortality largely a function of

That the relative incidence is more than a temporary accident is

1906-10 White Negro

40. Mental Achievement and Social

You might also like