1 The Internet of (Important) Things Thomas Watteyne 7 May 2019, Paris Presented in partial satisfaction of the requirements for the degree of “Habilitation à Diriger des Recherches” of Sorbonne University PhD • CITI, INSA Lyon • Orange Labs (CIFRE) 2009 2015 2017 Postdoc Prof. Kris Pister Project lead OpenWSN Sr. Networking Design Engineer IETF 6TiSCH co-chair • MEng Telecom, INSA Lyon (2005) • MSc, INSA Lyon (2005) 2011 2013 2019 Starting / Advanced Researcher Position Associate teams • UC Berkeley • Univ. Michigan • Univ. Southern California Falco co-founder 2/25
13
Embed
The Internet of (Important) Things - Thomas Watteyne...2019/05/19 · Power consumption Development cycles Node size 0% 20% 20% 60% 80% 100% source: OnWorld, 2005 1. Reliability:
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
The Internet of (Important) Things
Thomas Watteyne
7 May 2019, Paris
Presented in partial satisfaction of the requirements for the degree of
“Habilitation à Diriger des Recherches” of Sorbonne University
PhD
• CITI, INSA Lyon
• Orange Labs (CIFRE)
2009 2015 2017
Postdoc
Prof. Kris Pister
Project lead OpenWSN
Sr. Networking Design Engineer
IETF 6TiSCH
co-chair
• MEng Telecom, INSA Lyon (2005)
• MSc, INSA Lyon (2005)
2011 2013 2019
Starting / Advanced Researcher Position
Associate teams
• UC Berkeley
• Univ. Michigan
• Univ. Southern California
Falco
co-founder
2/25
2
• PhD students1. Keoma Brun-Laguna1
2. Jonathan Munoz2
3. Mina Rady1
• Postdocs1. Mališa Vučinić
2. Tengfei Chang
3. Ziran Zhang
4. Remy Leone
• Research Engineers1. Trifun Savic
2. Yasuyuki Tanaka1
• Undergraduate Interns1. Ba Hai Le
2. Felipe Moran
3. Fabian Rincon Vija
4. Marcelo Augusto Ferreira1 co-advised with Pascale Minet2 co-advised with Paul Muhlethaler
Grand ChallengeDependability“a network you can count on”
The Industrial Internet of Things (IIoT)
6/25
4
16 c
hannel offsets
e.g. 33 time slots
A
BC
DE
FG
H
I
J
• Motes are synchronized
• Communication follows a schedule
• Schedule gives tunable trade-off between
• packets/second
• latency
• robustness
…and energy consumption
7
Time Synchronized Channel Hopping
7/25
Approach
• minimal Viable Product (MVP)
• real-world validation
• cross-disciplinary research
• analysis
• simulation/emulation
• experimentation
• standardization
• Interop events
• benchmarkingTime
Synchronized
Channel
Hopping
8/25
5
SchedulingChannel Hopping Characterization
Representative Contributions
9/25
SchedulingChannel Hopping Characterization
Why Channel Hopping Makes Sense
1 T. Watteyne, A. Mehta, K. Pister., “Reliability Through Frequency Diversity: Why Channel Hopping Makes Sense“, ACM PE-WASUN, 2009.2 B. Kerkez, T. Watteyne, M. Magliocco, S. Glaser, K. Pister, “Feasibility Analysis of Controller Design for Adaptive Channel Hopping“, WSNPerf, 2009.3 J. Muñoz, P. Muhlethaler, X. Vilajosana, T. Watteyne. “Why Channel Hopping Makes Sense, even with IEEE802.15.4 OFDM at 2.4 GHz”. GIoTS, 2018.
• T. Watteyne, S. Lanzisera, A. Mehta, K. Pister, “Mitigating Multipath Fading Through Channel Hopping in Wireless Sensor Networks”, IEEE ICC, 2010.
• B. Kerkez, T. Watteyne, M. Magliocco, S. Glaser, K. Pister, “Feasibility Analysis of Controller Design for Adaptive Channel Hopping“, WSNPerf, 2009.
blind channel hopping
11/25
SchedulingChannel Hopping Characterization
• P. H. Gomes, T. Watteyne, B. Krishnamachari. “MABO-TSCH: Multi-hop And Blacklist-based Optimized Time Synchronized Channel Hopping”.
Wiley Transactions on Emerging Telecommunications (ETT), 2017.
Adaptive Channel Hopping using Game Theory
2.405 GHz 2.480 GHz
ACK=
1. ϵ-greedy algorithm
(here ϵ=0.025)
2. results embedded in ACK
(ordered list of frequencies)
“Multi-Armed Bandit” problem
37%
12/25
7
SchedulingChannel Hopping Characterization
A
BC
DE
FG
H
I
J
Trade off:
• Bandwidth
• Reliability
• Latency
Lifetime
Approaches:
• Centralized
• Distributed
1 M. Vučinić, T. Watteyne, X. Vilajosana. Broadcasting Strategies in 6TiSCH Networks. Wiley Internet Technology Letters, 2018.2 R. Rivest, Network Control by Bayesian Broadcast. IEEE Trans. Inform. Theory. 1987;33(3):323–328.3 T. Chang, M. Vucinic, X. Vilajosana, S. Duquennoy, D. Dujovne. 6TiSCH Minimal Scheduling Function (MSF). IETF [work-in-progress], 2018.
1 T. Chang, T. Watteyne, Q. Wang, X. Vilajosana. LLSF: Low Latency Scheduling Function for 6TiSCH Networks. IEEE DCOSS, 2016.
Aspect 2: Latency1
• Idea: cascade cell allocation
• Built into the 6top Protocol
A
BC
DE
FG
H
I
J
14/25
8
SchedulingChannel Hopping Characterization
1 M. Domingo-Prieto, T. Chang, X. Vilajosana, T. Watteyne. “Distributed PID-based Scheduling for 6TiSCH Networks”. IEEE Comm. Letters, 2016.
Aspect 3: Dynamic Resource Allocation1
A
BC
DE
FG
H
I
J
Reason 1: F starts
producing more data
Reason 2: PDR of
GE degrades
time
cell u
sage 100%
80%
remove cells
add cells
Proportional
Integral
Derivative
15/25
SchedulingChannel Hopping Characterization
Real-World Deploymentsover 1,000 sensors on 3 continents
gather
store and analyze visualize
exploring applicability through system-level and cross-disciplinary research
Mendoza, Argentina
Lorient, FranceCalifornia, USA
16/25
9
Machine learning:
• Random Forest
• K-Nearest-Neighbors
• Neural Network
• AdaBoost
SchedulingChannel Hopping Characterization
A Machine-Learning Based Connectivity Model
C. Oroza, Z. Zhang, T. Watteyne, S. Glaser, “Machine-Learning Based Connectivity Model for Complex Terrain Large-Scale Low-Power Wireless Deployments“,
IEEE Transactions on Cognitive Communications and Networking, 2017.
Goal: help deploy a network
predict connectivity
?
??
our 42,157,324 PDR
measurements
Annotations:
• Path ground distance
• Terrain complexity
• Vegetation variability
• Mean percent canopy
• Path angle
• Source canopy
• Receiver canopy
17/25
Transfer
Standardization
• Writing standards
• Interop events
• benchmarking
Deployments
Start-up
incubated at:
18/25
10
Wireless ControlAgile Networking Smart Dust
Research Program
19/25
OpenMote B
Agile Networking
Wireless ControlAgile Networking Smart Dust
13.2 km
100% PDR
RSSI -110 dBm
868 MHz
2-FSK@50-kbps
FEC
IEE
E8
02
.15
.4g
: 3
1 r
ad
io s
ett
ing
s
All setting, both 2.4
GHz and sub-GHz
A
BC
DE
F
G
H
I
J
IEEE802.15.4 PHY
IEEE802.15.4 MAC
6top [6P & SF]
IETF 6LoWPAN
IETF RPL
UDP
CoAP
• J. Munoz, T. Chang, X. Vilajosana, T. Watteyne, “Evaluation of IEEE802.15.4g for Environmental Observations”, MDPI Journal on Sensor Networks, 2018.
• J. Muñoz, P. Muhlethaler, X. Vilajosana, T. Watteyne, “Why Channel Hopping Makes Sense, even with IEEE802.15.4 OFDM at 2.4 GHz: GIoTS, 2018.
• J. Munoz, X. Vilajosana, T. Chang, “Problem Statement for Generalizing 6TiSCH to Multiple PHYs”, IETF 6TiSCH I-D [WIP], 2018.
1 SmartMesh Power and Performance Estimator, analog.com2 K. Brun-Laguna, “Deterministic Networking for the Industrial IoT”, PhD thesis, 2018.
SmartMesh performance estimator output1
Goal: generalized methodology to turn
(schedule+topology) into latency distribution
21/25
Wireless ControlAgile Networking Smart Dust
Control Loops
1 Schindler, Watteyne, Vilajosana, Pister, "Implem. and Charac. of a Multi-hop 6TiSCH Network for Exp. Feedback Control of an Inverted Pendulum: IEEE WiOpt, 2017.