Оглавление
Yang Sun Yang. Intelligent IoT for the Digital World
Table of Contents
List of Tables
List of Illustrations
Guide
Pages
Intelligent IoT for the Digital World. Incorporating 5G Communications and Fog/Edge Computing Technologies
Preface
Notes
Acknowledgments
Acronyms
1 IoT Technologies and Applications. 1.1 Introduction
1.2 Traditional IoT Technologies. 1.2.1 Traditional IoT System Architecture
1.2.1.1 Sensing Layer
1.2.1.2 Network Layer
1.2.1.3 Application Layer
1.2.2 IoT Connectivity Technologies and Protocols
1.2.2.1 Low‐power and Short‐range Connectivity Technologies. Radio Frequency IDentification
Near Field Communication
BLE
Zigbee
1.2.2.2 Low Data Rate and Wide‐area Connectivity Technologies. 6LoWPAN
LoRa
Sigfox
NB‐IoT
1.2.2.3 Emerging IoT Connectivity Technologies and Protocols. MIOTY
IoTivity
1.3 Intelligent IoT Technologies
1.3.1 Data Collection Technologies
1.3.1.1 mmWave
1.3.1.2 Massive MIMO
1.3.1.3 Software Defined Networks
1.3.1.4 Network Slicing
1.3.1.5 Time Sensitive Network
1.3.1.6 Multi‐user Access Control
1.3.1.7 Muti‐hop Routing Protocol
1.3.2 Computing Power Network. 1.3.2.1 Intelligent IoT Computing Architecture
1.3.2.2 Edge and Fog Computing
1.3.3 Intelligent Algorithms. 1.3.3.1 Big Data
1.3.3.2 Artificial Intelligence
1.4 Typical Applications. 1.4.1 Environmental Monitoring
1.4.2 Public Safety Surveillance
1.4.3 Military Communication
1.4.4 Intelligent Manufacturing and Interactive Design
1.4.5 Autonomous Driving and Vehicular Networks
1.5 Requirements and Challenges for Intelligent IoT Services. 1.5.1 A Generic and Flexible Multi‐tier Intelligence IoT Architecture
1.5.2 Lightweight Data Privacy Management in IoT Networks
1.5.3 Cross‐domain Resource Management for Intelligent IoT Services
1.5.4 Optimization of Service Function Placement, QoS, and Multi‐operator Network Sharing for Intelligent IoT Services
1.5.5 Data Time stamping and Clock Synchronization Services for Wide‐area IoT Systems
1.6 Conclusion
References
2.1 Introduction
2.2 Multi‐tier Computing Networks and Service Architecture
2.2.1 Multi‐tier Computing Network Architecture
2.2.2 Cost Aware Task Scheduling Framework
2.2.2.1 Hybrid Payment Model
2.2.2.2 Weighted Cost Function
Delay model
Energy consumption model
Weighted cost function
2.2.2.3 Distributed Task Scheduling Algorithm
Fog nodes broadcasting the workload status to UEs
UEs determining the best task processing mode
UE contending for the update opportunity of task processing mode
UE updating the task processing mode
2.2.3 Fog as a Service Technology
2.2.3.1 FA2ST Framework
2.2.3.2 FA2ST Application Deployment
2.2.3.3 FA2ST Application Management
2.3 Edge‐enabled Intelligence for Industrial IoT. 2.3.1 Introduction and Background. 2.3.1.1 Intelligent Industrial IoT
2.3.1.2 Edge Intelligence
2.3.1.3 Challenges
2.3.2 Boomerang Framework
2.3.2.1 Framework Overview
2.3.2.2 Prediction Model and Right‐sizing Model
2.3.2.3 Boomerang Optimizer
2.3.2.4 Boomerang with DRL
2.3.3 Performance Evaluation
2.3.3.1 Emulation Environment
2.3.3.2 Evaluation Results
2.4 Fog‐enabled Collaborative SLAM of Robot Swarm. 2.4.1 Introduction and Background
2.4.2 A Fog‐enabled Solution. 2.4.2.1 System Architecture
2.4.2.2 Practical Implementation
2.5 Conclusion
References
Notes
3 Cross‐Domain Resource Management Frameworks. 3.1 Introduction
3.2 Joint Computation and Communication Resource Management for Delay‐Sensitive Applications
3.2.1 2C Resource Management Framework. 3.2.1.1 System Model
3.2.1.2 Problem Formulation
Non‐splittable tasks and paired offloading
Splittable tasks and parallel offloading
3.2.2 Distributed Resource Management Algorithm. 3.2.2.1 Paired Offloading of Non‐splittable Tasks
Theorem 3.1
3.2.2.2 Parallel Offloading of Splittable Tasks
Algorithm 1 POMT Algorithm
Lemma 3.1
Lemma 3.2
Lemma 3.3
Theorem 3.2
Algorithm 2 POMT Algorithm
3.2.3 Delay Reduction Performance. 3.2.3.1 Price of Anarchy
Lemma 3.4
Lemma 3.5
3.2.3.2 System Average Delay
3.2.3.3 Number of Beneficial TNs
3.2.3.4 Convergence
3.3 Joint Computing, Communication, and Caching Resource Management for Energy‐efficient Applications
3.3.1 Fog‐enabled 3C Resource Management Framework
3.3.1.1 System Resources
IoT data cache
IoT devices and edge servers
3.3.1.2 Task Model
3.3.1.3 Task Execution
D2D input/downloading stage
Computation stage
Relaying and uploading stage
3.3.1.4 Problem Statement
Theorem 3.3
3.3.2 Fog‐enabled 3C Resource Management Algorithm. 3.3.2.1 F3C Algorithm Overview
3.3.2.2 F3C Algorithm for a Single Task
3.3.2.3 F3C Algorithm For Multiple Tasks
Definition 3.1
Definition 3.2
Definition 3.3
Permitted operations in task team formation
Definition 3.4
Definition 3.5
Definition 3.6
Definition 3.7
Task team formation algorithm
Algorithm 3 Initial Allocation Algorithm
Theorem 3.4
Algorithm 4 Task Team Formation Algorithm
3.3.3 Energy Saving Performance
3.3.3.1 Energy Saving Performance with Different Task Numbers
3.3.3.2 Energy Saving Performance with Different Device Numbers
3.4 Case Study: Energy‐efficient Resource Management in Tactile Internet
3.4.1 Fog‐enabled Tactile Internet Architecture
3.4.2 Response Time and Power Efficiency Trade‐off. 3.4.2.1 Response Time Analysis and Minimization
3.4.2.2 Trade‐off between Response Time and Power Efficiency
3.4.3 Cooperative Fog Computing. 3.4.3.1 Response Time Analysis for Cooperative Fog Computing with n FNs
3.4.3.2 Response Time and Power Efficiency Trade‐off for Cooperative Fog Computing Networks
3.4.4 Distributed Optimization for Cooperative Fog Computing
3.4.5 A City‐wide Deployment of Fog Computing‐supported Self‐driving Bus System. 3.4.5.1 Simulation Setup for Traffic Generated by a Self‐driving Bus
3.4.5.2 Simulation Setup for a Fog Computing Network
3.4.5.3 Numerical Results
3.5 Conclusion
References
Notes
4 Dynamic Service Provisioning Frameworks. 4.1 Online Orchestration of Cross‐edge Service Function Chaining. 4.1.1 Introduction
4.1.2 Related Work
4.1.3 System Model for Cross‐edge SFC Deployment
4.1.3.1 Overview of the Cross‐edge System
4.1.3.2 Optimization Space
4.1.3.3 Cost Structure
4.1.3.4 The Cost Minimization Problem
4.1.4 Online Optimization for Long‐term Cost Minimization
4.1.4.1 Problem Decomposition via Relaxation and Regularization
Algorithm 5 An online regularization-based fractional algorithm – ORFA
4.1.4.2 A Randomized Dependent Rounding Scheme
Algorithm 6 Randomized dependent instance provision – RDIP
4.1.4.3 Traffic Re‐routing
4.1.5 Performance Analysis
4.1.5.1 The Basic Idea
4.1.5.2 Competitive Ratio of ORFA
Theorem 4.1
4.1.5.3 Rounding Gap of RDIP
Theorem 4.2
4.1.5.4 The Overall Competitive Ratio
Theorem 4.3
4.1.6 Performance Evaluation
4.1.6.1 Experimental Setup
4.1.6.2 Evaluation Results
4.1.7 Future Directions
4.2 Dynamic Network Slicing for High‐quality Services
4.2.1 Service and User Requirements
4.2.2 Related Work
4.2.3 System Model and Problem Formulation. 4.2.3.1 Fog Computing
4.2.3.2 Existing Network Slicing Architectures
4.2.3.3 Regional SDN‐based Orchestrator
4.2.3.4 Problem Formulation
4.2.4 Implementation and Numerical Results. 4.2.4.1 Dynamic Network Slicing in 5G Networks
4.2.4.2 Numerical Results
4.3 Collaboration of Multiple Network Operators
4.3.1 Service and User Requirements
4.3.2 System Model and Problem Formulation. 4.3.2.1 IoT Solutions in 3GPP Release 13
4.3.2.2 Challenges for Massive IoT Deployment
4.3.2.3 Inter‐operator Network Sharing Architecture
4.3.2.4 Design Issues
4.3.3 Performance Analysis
4.4 Conclusion
References
Notes
5 Lightweight Privacy‐Preserving Learning Schemes* 5.1 Introduction
5.2 System Model and Problem Formulation
5.3 Solutions and Results
5.3.1 A Lightweight Privacy‐preserving Collaborative Learning Scheme
5.3.1.1 Random Gaussian Projection (GRP)
Property 5.1
Property 5.2
5.3.1.2 Gaussian Random Projection Approach
5.3.1.3 Illustrating Examples
5.3.1.4 Evaluation Methodology and Datasets
5.3.1.5 Evaluation Results with the MNIST Dataset
5.3.1.6 Evaluation Results with a Spambase Dataset
5.3.1.7 Summary
5.3.1.8 Implementation and Benchmark
5.3.2 A Differentially Private Collaborative Learning Scheme
5.3.2.1 Approach Overview
5.3.2.2 Achieving ε‐Differential Privacy
5.3.2.3 Performance Evaluation
5.3.3 A Lightweight and Unobtrusive Data Obfuscation Scheme for Remote Inference
5.3.3.1 Approach Overview
5.3.3.2 Case Study 1: Free Spoken Digit (FSD) Recognition
5.3.3.3 Case Study 2: Handwritten Digit (MNIST) Recognition
5.3.3.4 Case Study 3: American Sign Language (ASL) Recognition
5.3.3.5 Implementation and Benchmark
5.4 Conclusion
References
Note
6 Clock Synchronization for Wide‐area Applications1. 6.1 Introduction
6.2 System Model and Problem Formulation
6.2.1 Natural Timestamping for Wireless IoT Devices
6.2.2 Clock Synchronization for Wearable IoT Devices
6.3 Natural Timestamps in Powerline Electromagnetic Radiation
6.3.1 Electrical Network Frequency Fluctuations and Powerline Electromagnetic Radiation
6.3.2 Electromagnetic Radiation‐based Natural Timestamping
6.3.2.1 Hardware
RPi‐based EMR sensor
Z1 mote
ENF database node
6.3.2.2 ENF Extraction
Challenges of zero crossing detection
Band‐pass filtering
6.3.2.3 Natural Timestamp and Decoding
Definition 6.1 (Decoding condition C1)
6.3.3 Implementation and Benchmark
6.3.3.1 Timestamping Parameter Settings
6.3.3.2 Z1 Implementation and Benchmark
6.3.4 Evaluation in Office and Residential Environments
6.3.4.1 Deployment and Settings
6.3.4.2 Evaluation Results. Performance evaluation at various sites
One‐month evaluation
Impact of deployment position
Storage volume constraint
System overhead
6.3.5 Evaluation in a Factory Environment
6.3.6 Applications
6.3.6.1 Time Recovery
6.3.6.2 Run‐time Clock Verification
6.3.6.3 Secure Clock Synchronization
Susceptibility of natural timestamping to EMR spoofing
Attack model and secure clock synchronization
Definition 6.2 (Packet delay attack)
6.4 Wearables Clock Synchronization Using Skin Electric Potentials
6.4.1 Motivation
6.4.2 Measurement Study
6.4.2.1 Measurement Setup
6.4.2.2 iSEP Sensing under Various Settings
Shared ground
Independent grounds
Summary
6.4.3 TouchSync System Design
6.4.3.1 TouchSync Workflow
6.4.3.2 iSEP Signal Processing
Band‐pass filter (BPF) or Mean Removal Filter (MRF)
Zero Crossing Detector (ZCD)
Phase‐locked Loop (PLL)
6.4.3.3 NTP Assisted with Dirac Combs
Protocol for a synchronization session
Clock offset analysis
Integer Ambiguity Solver (IAS)
An example of a unique solution
Program for solving integer ambiguity
Convergence speed
Algorithm 7 Slave's pseudocode for a synchronization process3
Algorithm 8 Master's pseudocode for a synchronization process
Discussions
6.4.4 TouchSync with Internal Periodic Signal
6.4.4.1 Extended Design
Protocol for IPS‐based TouchSync
Adaptive period mechanism (APM)
6.4.4.2 Numeric Experiments
Distribution of clock offset estimation errors
Convergence speed
6.4.4.3 Impact of Clock Skews
6.4.5 Implementation
6.4.6 Evaluation
6.4.6.1 Signal Strength and Wearing Position
6.4.6.2 Impact of High‐Power Appliances on TouchSync
6.4.6.3 Evaluation in Various Environments
Laboratory
Home
Office
Corridor
6.4.6.4 TouchSync‐over‐internet
6.5 Conclusion
References
Notes
7 Conclusion
Index. a
b
c
d
e
f
g
h
i
k
l
m
n
o
p
q
r
s
t
u
v
w
z
WILEY END USER LICENSE AGREEMENT
