Automating Google Workspace (GWS) & more with Apps Script
Requirements and Challenges of Smart Mobility for IoT
1. Mobile Wireless Systems for
Smart Mobility
Ana Aguiar
ana.aguiar@fe.up.pt
Universidade do Porto
Instituto de Telecomunicações
CooDriver GmbH
2. Change the way people move
Improve Mobility
Experience
Better
Planning
Internet of Things
Smart Mobility
3. Smart Mobility: Requirements
Better Planning Better Experience
Usage • Better understanding of
mobility processes and user
behavior and decisions
• More granular spatio-
temporal historic data
• Monitoring policy impact
• Services to inform decision
making: routing, driver
assistance systems
• Infotainment
• Traffic management
• Safety
• Autonomous driving
Actuation • Medium and long term
policy
• Interactive services
• Automation
Network • No real-time
• Big data, cloud computing
• Data warehousing
• Real-time
• Network management
• Computation distribution
User • People, (mostly) non-
technical
• People
• Machines
4. SenseMyCity: a Mobile IoT Crowdsensor
Leverage the power of the crowd to
sense large-scale human processes
J. Rodrigues, A. Aguiar, J. Barros. SenseMyCity: Crowdsourcing an Urban Sensor, arXiv:1412.2070, 2014
5. Campus Mobility Sustainability Study
Joint work with Cecília Silva, CITTA/ FEUP
If all trips within a real
distance of 3km were made
on foot or by bike, footprint
would go down 14%.
Active Modes
Ana Cláudia Proença, João Teixeira, Cecília Silva. Conferência AESOP, Lisboa, Jul 2017
Total: 150
Regular: 79
Total: 239
Regular: 150
2016 2017
Population in FEUP Campus: > 8000 people
Geographic distribution
of transportation modes
6. Data Quality: Impact on Extracted Information
J. Rodrigues, J. Pereira and A. Aguiar. "Impact of Crowdsourced Data Quality on Travel Pattern Estimation",
1st ACM Workshop on Mobile Crowdsensing Systems and Applications (CrowdSenSys 17), 2017
Average trip distance
Transportation mode
detection accuracy
7. Data Quality: Crowdsensed Data for Traffic Estimation
Population bias
D. Socas Gil, P. M. d'Orey, A. Aguiar. 2017. On the Challenges of Mobile Crowdsensing for Traffic
Estimation. ACM Conference on Embedded Network Sensor Systems (SenSys '17).
Sparsity, imbalance
0%
1%
2%
3%
4%
5%
6%
7%
8%
9%
10%
0 3 6 9 12 15 18 21 24
Percentageofsessions
Hour of the day
SenseMyCity SenseMyMood SenseMyFEUP
J GP Rodrigues, A. Aguiar, C. Queirós. Opportunistic Mobile Crowdsensing as a Transportation Systems
Tool, in Proc. 19th IEEE Intelligent Transportation Systems Conference (ITSC). 2016.
8. Smart Mobility: Challenges
Better Planning
• Key Performance Indicators’s
interdisciplinary nature
– What to collect and store?
– Calibration/ validation
• Sensing and data collection
– Obtaining representative data
– Privacy can (and must) be
addressed, as data is useful as an
aggregate
• Data quality and openness
– Law of large numbers can help
– Lack of standards for dataset pre-
processing, description or quality
– Data security: integrity
• Data quality
– Data inaccuracy, bias, etc
• Visualisation
9. Smart Mobility: Requirements
Better Planning Better Experience
Usage • Better understanding of
mobility processes and user
behavior and decisions
• More granular spatio-
temporal historic data
• Monitoring policy impact
• Services to inform decision
making: routing, driver
assistance systems
• Infotainment
• Traffic management
• Autonomous driving
• Safety
Actuation • Medium and long term
policy
• Interactive services
• Automation
Network &
Computing
• No real-time
• Big data, cloud computing
• Data warehousing
• Real-time
• Network management
• Computation distribution
User • People, (mostly) non-
technical
• People
• Machines
10.
11. Safeguarding internet experience of road-side mobile users
Reduce load on mobile hotspots
Prevent Short-lived Connections
to Vehicular Hotspots
L. Kholkine, P. Santos, A. Cardote, A. Aguiar, Detecting Relative Position of User Devices and Mobile Access Points,
IEEE Vehicular Networking Conference VNC, Columbus, United States, December, 2016.
Is this really a problem?
12. Prevent Short-lived Connections
to Vehicular Hotspots
• Identify user inside/outside bus via
classifier trained with large-scale
real-world dataset
• Implemented connection access in
WiFi association flow
• Experimental results: unnecessary
(outside) connections on vehicular
hotspot reduced by 40%
P. M. Santos, L. Kholkine, A. Cardote, A. Aguiar.
Context classifier for position-based user association control in vehicular hotspots,
Elsevier Computer Communications, Volume 121, Pages 71-82, 2018.
13. Bike2X: Impact of Antenna Position for 2.4GHz
Back rack
Under seat
Chain stay (chain side)
Chain stay (no chain)Handlebar
Frame Diamond
Handlebar
Chain stay
(chain side)
Chain stay
(no chain)
Back
rack
Under
seat
Frame
Diamond
Prx [dBm] = Ptx + GB-A (tx) + Lpl + GB-A (rx) + X(0,s)
P M Santos, L Pinto, A Aguiar, L Almeida. Characterization
and Modeling of the Bicycle-Antenna System for the 2.4GHz
ISM Band, Proc. IEEE Vehicular Networking Conference (VNC)
Taipei, Taiwan. 2018.
No significant impact of material observed
14. Bike2X: Communication to in-Car Hotspot
Parked car
Approaching bicycle
Pedro M. d'Orey, José Pintor, Pedro Miguel Salgueiro dos Santos, Ana Aguiar, Opportunistic Use of In-Vehicle
Wireless Networks for Vulnerable Road User Interaction, 2019 IEEE Intelligent Vehicles Symposium (IV2019),
Paris, France. June 2019.
Stopped car
Approaching bicycle
200 150 100 50 0 50 100 150 200
Tx-Rx distance (m)
1.0
2.0
5.5
11.0
12.0
18.0
24.0
36.0
48.0
54.0
DataRate(Mbit/s)
Approaching car left
Departing car right
Approaching car right
Departing car left
Bike antenna on
handlebar
15. Impact of Human Mobility in Mobile Offloading
High variability among users -> per-user basis offloading
decision
Tthresh={10,5} sec ~ similar availability of ORs to the
users:
Tthresh = 5 sec halves the transition time when compared
to Tthresh = 10 sec;
ORs are more likely to belong to the low-relevance
category -> Need for opportunistic offloading decision
models
E. Lima, A. Aguiar, A. Viana and P. Carvalho. Impacts of Human Mobility in Data Offloading, ACM MobiCom
Workshop on Challenged Networks (CHANTS), New Delhi, India. 2018.
16. IoT Platform Service Chain Delays
0.3130s ± 0.0470
M2MGWs
Data
ProcessorNSCL openEHR
Event
A
A
Time Time Time Time
A
K
K
A
K
K
A
K
A
K
K
A
A
K
K
A
OWDNSCLmId
K,A
OWDDP
A,K
RTTGW,NSCL
A
OWDGW,NSCL
A
OWDDP
K,A
OWDDP,EHR
A
OWDNSCL,EHRA
OWD
DP,NSCL
A
A
OW
D NSCL,DP
OWDNSCLmIa
K,A
OWDNSCLmIa
A,K
OWDNSCLmIa
A,K
0.0035s ± 1.162 e-4
0.1076s ± 0.0306
0.1399s ± 0.404 e-4
0.0138s ± 1.802e-4
0.0074s ± 7.535 e-5
C. Pereira, A. Pinto, D. Ferreira, A. Aguiar.
Experimental Characterisation of Mobile IoT Application Latency.
IEEE Internet of Things Journal, 2017.
17. Smart Mobility: Challenges
Better Planning
• Key Performance Indicators’s
interdisciplinary nature
– What to collect and store?
• Sensing and data collection
– Obtaining representative data,
calibration/ validation
– Privacy can be addressed, as data is
useful as an aggregate
• Data quality and openness
– Law of large numbers can help
– Lack of standards for dataset pre-
processing, description or quality
– Data security: integrity
• Data quality
– Data inaccuracy, bias, etc
• Visualisation
Better Experience
• Collect, process and manage large
amounts of small data pieces in real-
time
• Full system integration
• Low end-to-end virtualized service
chain latency
– Edge computing, SDN + NFV
• Leveraging all available spectrum
– Licensed, unlicensed, d2d (v2v)
• Computation distribution
• Data quality
– Positioning quality: defining needs
• Privacy and trust
Resources
• People
• Time
Thank you!!