Intelligent IoT for the Digital World. Yang Sun Yang

The architecture of an RFID system.Figure 1.2 NFC system architecture.Figure 1.3 Zigbee system architecture.Figure 1.4 6LoWPAN header layout.Figure 1.5 A BLE network.Figure 1.6 LoRa network architecture.Figure 1.7 Sigfox network architecture.Figure 1.8 NB‐IoT network architecture.Figure 1.9 TS‐UWB telegram splitting.Figure 1.10 OCF core framework.Figure 1.11 The number of connected devices worldwide 2015–2025.Figure 1.12 A traditional cloud‐based IoT architecture.Figure 1.13 A generic and flexible multi‐tier user‐centered IoT network arch...Figure 1.14 Application scenarios of intelligent IoT.Figure 1.15 Network slices existing in 5G architecture.Figure 1.16 The architecture of intelligent IoT multi‐tier computing.Figure 1.17 The network architecture of public safety surveillance.Figure 1.18 Intelligent transportation system scenario overview.Figure 1.19 Description of VANET for autonomous vehicles.

      2 Chapter 2Figure 2.1 A multi‐tier computing network architecture. The architecture int...Figure 2.2 An illustration of multi‐tier computing networks (Liu et al., 202...Figure 2.3 The cloud–edge–things system (left), the FA2ST system (right) (Ch...Figure 2.4 The FA2ST framework (Chen et al., 2018a).Figure 2.5 FA2ST application deployment (Chen et al., 2018a).Figure 2.6 Example scenario: edge intelligence enabled product surface inspe...Figure 2.7 Major edge‐centric inference modes: edge‐based, device‐based, edg...Figure 2.8 DNN partitioning and DNN right‐sizing.Figure 2.9 The workflow of Boomerang.Figure 2.10 Point selection algorithm in the Boomerang optimizer.Figure 2.11 The workflow of the DRL model for point selection.Figure 2.12 Evaluation results under different bandwidths and different late...Figure 2.13 The illustration of fog‐enabled robot SLAM system architecture (...Figure 2.14 The workflow of fog‐enabled robot SLAM (Yang et al., 2020b).Figure 2.15 The function view of fog‐enabled robot SLAM (Yang et al., 2020b)...Figure 2.16 The developed demo of fog‐enabled robot SLAM with OpenLTE. (a) m...Figure 2.17 The function view of fog‐enabled robot SLAM (Yang et al., 2020b)...Figure 2.18 The function view of fog‐enabled robot SLAM (Yang et al., 2020b)...Figure 2.19 Challenges and requirements solved and satisfied by the fog‐enab...

      3 Chapter 3Figure 3.1 A general heterogeneous MTMH fog network. The FNs have different ...Figure 3.2 System average delay with different number of TNs (non‐splittable...Figure 3.3 System average delay with different number of TNs, under differen...Figure 3.4 System average delay with different number of TNs, under differen...Figure 3.5 Number of beneficial TNs with different number of TNs (non‐splitt...Figure 3.6 Number of beneficial TNs with different number of TNs (splittable...Figure 3.7 Number of decision slots with different number of TNs.Figure 3.8 An illustration of collaborative task execution in fog computing....Figure 3.9 Cooperative F3C task execution.Figure 3.10 1C resource sharing example.Figure 3.11 2C resource sharing example.Figure 3.12 Basic procedures of the F3C algorithm.Figure 3.13 Minimum cost flow network transformation in single task. (a) Coo...Figure 3.14 An auxiliary graph

of minimum cost flow for task t.Figure 3.15 Performance gain in energy with different task numbers.Figure 3.16 Performance gain in energy with different device numbers.Figure 3.17 Participated device number under increasing device numbers.Figure 3.18 Average permitted operation number under different task numbers....Figure 3.19 Average permitted operation number under different FN numbers.Figure 3.20 Fog computing‐supported tactile internet architecture.Figure 3.21 (a) Response time under different amounts of workload processed ...Figure 3.22 Response time under different amounts of processed workload and ...Figure 3.23 (a) Distribution of FNs, bus routes, and considered areas, (b) d...Figure 3.24 Average workload (number of requests) processed by each FN at di...Figure 3.25 Average workload (number of requests) processed by each FN with ...Figure 3.26 Average response time of users with different power efficiency....

      4 Chapter 4Figure 4.1 An example of three service function chains (SFCs) for live video...Figure 4.2 An illustration of cross‐edge service function chain deployment....Figure 4.3 An illustration of the basic idea of the performance analysis.Figure 4.4 The workload trace of Google, Facebook, HP, and Microsoft data ce...Figure 4.5 The competitive ratio of different online algorithms.Figure 4.6 The effect of the switching cost on the competitive ratio.Figure 4.7 The effect of the control parameter ε on the competitive rat...Figure 4.8 The competitive ratio of various rounding schemes.Figure 4.9 The overall competitive ratio of different algorithm combinations...Figure 4.10 Dynamic network slicing architecture.Figure 4.11 Distribution of BSs and considered areas.Figure 4.12 Offloaded workload for both types of service in different areas ...Figure 4.13 Offloaded workload under different RTT between fog nodes.Figure 4.14 Offloaded workload under different workload arrival rates for im...Figure 4.15 Offloaded workload under different amounts of harvested energy....Figure 4.16 Inter‐operator network sharing: (a) a spectrum pool and (b) spec...Figure 4.17 Locations of BSs deployed by two major cellular operators in the...Figure 4.18 Maximum number of IoT (eMTC) devices that can coexist with cellu...Figure 4.19 Maximum number of IoT (eMTC) devices that can coexist with cellu...

      5 Chapter 5Figure 5.1 A collaborative learning system.Figure 5.2 Two‐dimensional example. Original data vectors and projected data...Figure 5.3 Test accuracy based on projected data versus the number of partic...Figure 5.4 Test accuracy based on projected data versus the condition number...Figure 5.5 Example images from the MNIST dataset.Figure 5.6 CNN with a projected MNIST image as input.Figure 5.7 Impact of the number of participants (MNIST). The error bars for ...Figure 5.8 Impact of data compression on learning performance (MNIST,

).Figure 5.9 Impact of differential privacy loss on learning performance (MNIS...Figure 5.10 Impact of the number of participants (spambase). The error bars ...Figure 5.11 Overview of our proposed privacy‐preserving collaborative learni...Figure 5.12 CNN structure.Figure 5.13 Impact of privacy loss level ε on the test accuracy of the ...Figure 5.14 Impact of batch size on the test accuracy of the collaboratively...Figure 5.15 Impact of privacy loss level ε on the test accuracy of the ...Figure 5.16 ObfNet for remote inference. The fog node i desires privacy prot...Figure 5.17 The procedure to generate ObfNets.Figure 5.18 Structure of

for FSD recognition.Figure 5.19 Structure of

for FSD recognition.Figure 5.20 Test accuracy of different ObfNet–InfNet concatenations in 10 te...Figure 5.21 Structure of

for MNIST recognition.Figure 5.22 Structure of

for MNIST recognition.Figure 5.23 Test accuracy of InfNets and ObfNet–InfNet concatenations for MN...Figure 5.24 Obfuscation results of ObfNet

on MNIST.Figure 5.25 Structure of