Evaluation of Energy-efficient Data Collection in

GreenDays@Sophia 2017

Evaluation of Energy-efficient Data Collection in Mobile Crowdsensing Systems with CrowdSenSim Simulator Andrea Capponi

University of Luxembourg

Claudio Fiandrino

Imdea Networks Institute

Giuseppe Cacciatore Dzmitry Kliazovich Pascal Bouvry

University of Trento ExaMotive University of Luxembourg June 27, 2017

Smart Cities: Introduction I

50% of worldwide population lives in cities

I

Cities account for I

80% of worldwide gas consumption

I

75% of global energy consumption

I

60% of residential water use

N. B. Grimm, S. H. Faeth, N. E. Golubiewski, C. L. Redman, J. Wu, X. Bai, and J. M. Briggs, “Global change and the ecology of cities,” in Science, vol. 319, no. 5864, 2008, pp. 756–760.

Andrea Capponi | GreenDays@Sophia 2017 | Evaluation of Energy-efficient Data Collection in MCS Systems with CrowdSenSim

1 of 28

IoT for Smart Cities Improve citizens’ quality of life with innovative and sustainable solutions for public services: I

Smart parking

I

Waste management

I

Public lighting

I

IoT candidate building block for Smart Cities solutions


2 of 28

Sensing as a Service (S2 aaS) for IoT Sensors

Consumers

Cloud Providers


3 of 28

Mobile Crowdsensing I

Appealing paradigm for sensing and collecting data I

Monitoring phenomena in smart cities

I

Sensing as a Service (S2 aaS) business model

I

Sensors commonly available in mobile and IoT devices


4 of 28




I


I


Figure: Diffusion of mobile devices Andrea Capponi | GreenDays@Sophia 2017 | Evaluation of Energy-efficient Data Collection in MCS Systems with CrowdSenSim

4 of 28




I


I


Figure: Diffusion of mobile devices Andrea Capponi | GreenDays@Sophia 2017 | Evaluation of Energy-efficient Data Collection in MCS Systems with CrowdSenSim

4 of 28

Sensing Paradigms

I

Two sensing paradigms:

Participatory

Opportunistic

I

Active user engagement

I

Minimal user involvement

I

Sensing and reporting user-driven

I

Sensing and reporting application- or device-driven

I

Centralized, direct task assignment

I

Distributed, no direct task assignment


5 of 28

Outline

Framework for Data Collection in Opportunistic Sensing Systems

User Recruitment

Smart Lighting


6 of 28

Outline


User Recruitment

Smart Lighting


6 of 28

Mobile Crowdsensing Scenario Environmental Context Cs

S2 aaS Cloud Collector

Crowd iFi W

ta Da

Data Collection Utility dsa

E LT

Threshold δ

Accelerometer

IoT Devices Gyroscope

Microphone

Dual Camera

Temperature

Smartphone Sensing Potential sps


7 of 28

Mobile Crowdsensing Scenario Environmental Context Cs

S2 aaS Cloud Collector

Crowd iFi W

ta Da

Data Collection Utility dsa

E LT

Threshold δ

Accelerometer

IoT Devices Gyroscope

Microphone

Dual Camera

Temperature

Smartphone Sensing Potential sps


7 of 28

Proposed Opportunistic Framework

Sensing Decision Policy Cs · [γ · sps + (1 − γ) · dsa ] > δ


8 of 28


Data Collection Utility

Threshold

Sensing Decision Policy Cs · [γ · sps + (1 − γ) · dsa ] > δ

Environmental Context

Smartphone Sensing Potential

Balancing Coefficient


8 of 28


I

Data Collection Utility dsa I I

I

Defines utility for the cloud collector Based on number of samples already received

Smartphone Sensing Potential, sps I

Function of locally spent energy Es for sensing and reporting Es = Esc + Esr

I

Environmental Context Cs I

I

Binary value which depends on the context awareness

Threshold δ I

Depends on the level of battery of the devices (B) and the amount of reported data that devices have already contributed (D)


9 of 28

CrowdSenSim I

Custom simulator for crowdsensing activities I I

Access and download: http://crowdsensim.gforge.uni.lu Contact: [email protected]


10 of 28

CrowdSenSim: Features and Architecture I

Large scale (time-space)

I

Realistic urban environments City Layout

List of Events

CrowdSensing Inputs

User Mobility

S IMULATOR

Results


11 of 28

Evaluation Settings

I

3 sensors: accelerometer, temperature, pressure

I

Users: 20 000

I

Walking speed: U [1 − 1.5] m/s

I

Walking period: U [10 − 20] min

I

Simulation period: 8 AM - 2 PM

I

Digipoint to provide street-level maps

I

City centers of Luxembourg, Madrid and Trento


12 of 28

Performance Evaluation

Traces

0-50 50-100 100-150 150-200 200-250 250-300 300-350 350-400 400-450 450-500 500-550 550-600 600-650 650-700 700-750 750-800 800-850 850-900

Uniform

Current drain (µAh)

(a) Sensing Cost

3 500 3 000 2 500 2 000 1 500 1 000 500 0

Uniform

Traces

0-5 5-10 10-15 15-20 20-25 25-30 30-35 35-40 40-45 45-50 50-55 55-60 60-65 65-70 70-75 75-80 80-85

3 500 3 000 2 500 2 000 1 500 1 000 500 0

Num. of Participants

Energy spent for sensing and communications

Num. of Participants

I

Energy (J)

(b) Communication Cost

I

Sensing and communication same distribution but different scales

I

Real-world traces are the results of a study on pedestrian mobility and are public available on Crawdad (ostermalm_dense_run2) 1 1 http://crawdad.org/kth/walkers/20140505


13 of 28

Performance Evaluation

I

Normalized distribution of amount of collected data for the different cities over the time period 8:00 AM - 2:00 PM

0

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

(a) Luxembourg

1

0

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

(b) Trento

1

0

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

1

(c) Madrid


14 of 28

Outline


User Recruitment

Smart Lighting


14 of 28

The DSE Recruitment Policy The recruitment problem

I

I

Select users able to fulfill the task with high accuracy

I

Minimize costs of the organizer/users

I

Recruit a minimum number of users N

Three factors define user eligibility I I I

I

Distance from sensing task Sociability Remaining battery charge

Acceptance: sociability and energy


15 of 28

Cloud

High Computing Capacity High Latency Global View

O RGANIZER /C OLLECTOR: Global Analytics, Task Definition, Initiator User Recruitment Process

Fog

Proposed Fog Platform

C LOUDLET: Local Analytics, ID assignment, Computation of Recruitment Factor

Low Computing Capacity Low Latency Local View

I

Fog computes social factor

I

Extension: anonymity, privacy

M OBILE I O T D EVICES: Sensing Task


16 of 28

Evaluation Settings

I

10 000 participants

I

6 cloudlets

I


I


I

Simulation period: 8 AM - 2 PM

I

Set of 25 tasks (40 minutes each)


17 of 28

Distribution of Contacted vs Recruited Users I

DSE policy

Contacted

Recruited

Number of Users

150 125 100 75 50 25 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Task ID Andrea Capponi | GreenDays@Sophia 2017 | Evaluation of Energy-efficient Data Collection in MCS Systems with CrowdSenSim

18 of 28

Distribution of Contacted vs Recruited Users I

Distance-only policy

Contacted

Recruited

Number of Users

150 125 100 75 50 25 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Task ID Andrea Capponi | GreenDays@Sophia 2017 | Evaluation of Energy-efficient Data Collection in MCS Systems with CrowdSenSim

18 of 28

Outline


User Recruitment

Smart Lighting


19 of 28

Current Street Lighting Implementations (I) Not environmental sustainable: I

19% of worldwide use of electrical energy

I

6% of total emissions of greenhouse gases

I

40% of the cities’ energy budget

European Energy Innovation Publisher, “LED Lighting Technologies”, in 2nd Annual LED Professional Symposium & Exhibition.


20 of 28

Current Street Lighting Implementations (II)

Not energy efficient because of I

technology

I

control system

T YPE OF L AMPS

N OM . WATTAGE (W)

L AMP E FFICACY ( LM /W)

E NERGY ( K W H /1000 H )

AVERAGE LIFE ( H )

HPS-97241 HPS-93010296 MH-NaSc LED-GRN60 LED-GRN100

150.0 250.0 100.0 46.8 73.3

110.0 129.0 90.0 131.0 138.0

172.7 283.4 165.0 51.8 82.7

24 000 24 000 10 000 100 000 100 000


21 of 28

New Heuristics for Street Lighting (I) The new proposed heuristics rely on occupancy 2 and each lamppost operates independently I

IoT-augmented lamps

I

Strategies: I

Delay-based (D EL)

I

Encounter-based (E NC)

I

Dimming (D IM)

R

2 F. Leccese, "Remote-Control System of High Efficiency and Intelligent Street Lighting Using a ZigBee Network of Devices and Sensors," in IEEE Transactions on Power Delivery, vol. 28, no. 1, pp. 21-28, Jan. 2013. Andrea Capponi | GreenDays@Sophia 2017 | Evaluation of Energy-efficient Data Collection in MCS Systems with CrowdSenSim

22 of 28

New Heuristics for Street Lighting (II) M ETHOD

D ESCRIPTION

Current (C UR)

Lampposts continuously active emitting maximum light intensity

LO

Delay-based (D EL)

Lampposts switched on when users pass nearby. If nobody is present within R, lampposts remain active for time window W and then are switched off.

HI

Encounter-based (E NC)

Lampposts switched upon the first encounter with at least one user and remain active the whole night.

ME

Dimming (D IM)

Lampposts operate at 60% light intensity in absence of users within R. Lampposts light up/dim the intensity in proportion to the number of nearby users.

HI


E FFICACY

23 of 28

Evaluation Settings

Luxembourg City center

I

Users: 20 000

I


I


I

Simulation period: 9 PM - 7 AM

I

537 Lampposts

I

R: {10, 30, 50} m

0.3

PDF

I

0.2 0.1 0

9-10 10-11 11-12 12-1 1-2 PM

2-3

3-4

4-5

5-6

6-7

AM

Time Interval (hour)


24 of 28

Energy consumption (kW/day)

Energy Comparison of Heuristics C UR

E NC

D EL

D IM

1 000 800 600 400 200 0

10

30

50

Radius (m)

I

D IM significantly outperforms other heuristics

I

E NC already improves current systems


25 of 28

Heatmap of Energy Consumption

I

Comparison ENC vs DEL3 170

170

160

160

150

150

140

140

130

130

120

120

110

110

100

100

90

90

80

80

70

70

60

60

50

50

3 https://developers.google.com/maps/documentation/javascript/examples/layerheatmap Andrea Capponi | GreenDays@Sophia 2017 | Evaluation of Energy-efficient Data Collection in MCS Systems with CrowdSenSim

26 of 28

Conclusion CrowdSenSim I

Performance evaluation in large realistic urban environments I

MCS systems for S2 aaS applications in smart cities I I I

I

Framework for cost effective data collection Energy efficient user recruitment Energy efficient public street lighting

Custom simulator for crowdsensing activities I

Access and download: http://crowdsensim.gforge.uni.lu

I

Contact: [email protected]

Future work I

HyDACQ: Hybryd Data ACQuisition paradigm

I

Smartphone cooperation for sensing and data relaying using D2D More advanced solutions: predicted users’ mobility, cluster control

I


27 of 28

Publications 1. A. Capponi, C. Fiandrino, C. Franck, U. Sorger, D. Kliazovich, and P. Bouvry, "Assessing Performance of Internet of Things-Based Mobile Crowdsensing Systems for Sensing as a Service Applications in Smart Cities", in IEEE CLOUDCOM, Dec 2016, Luxembourg. 2. C. Fiandrino, A. Capponi, G. Cacciatore, D. Kliazovich, U. Sorger, P. Bouvry, B. Kantarci, F. Granelli and S. Giordano, "CrowdSenSim: a Simulation Platform for Mobile Crowdsensing in Realistic Urban Environments", in IEEE ACCESS, Feb 2017. 3. A. Capponi, C. Fiandrino, D. Kliazovich, P. Bouvry and S. Giordano, "Energy Efficient Data Collection in Opportunistic Mobile Crowdsensing Architectures for Smart Cities", in IEEE INFOCOM WKSHPS on Smart Cities and Urban Computing (SmartCity), May 2017, Atlanta, GA, USA. 4. A. Capponi, C. Fiandrino, D. Kliazovich, P. Bouvry and S. Giordano, "A Cost-Effective Distributed Framework for Data Collection in Cloud-based Mobile Crowd Sensing Architectures", in IEEE Transactions on Sustainable Computing, Mar 2017, 2 (1), pp. 3-16. ISSN: 2377-3782, DOI: 10.1109/TSUSC.2017.2666043. 5. C. Fiandrino, F. Anjomshoa, B. Kantarci, D. Kliazovich, P. Bouvry, J. Matthews, "Sociability-Driven Framework for Data Acquisition in Mobile Crowdsensing over Fog Computing Platforms for Smart Cities," in IEEE Transactions on Sustainable Computing, May 2017, doi: 10.1109/TSUSC.2017.2702060. 6. G. Cacciatore, C. Fiandrino, D. Kliazovich, P. Bouvry, and F. Granelli, "Cost Analysis of Smart Lighting Solutions for Smart Cities", in IEEE Intenational Conference on Communications (ICC), May 2017, Paris, France. Andrea Capponi | GreenDays@Sophia 2017 | Evaluation of Energy-efficient Data Collection in MCS Systems with CrowdSenSim

28 of 28

Thank You! Andrea Capponi

Data Collection Utility, dsa

I

Based on number of samples already received

I

Can be defined as the following sigmoid function: dsa =

I I

1 1+e

s −ϕ ρs

a

·(−N s |t +(1− ρ2s ))

ϕs and (1 − ρs /2) coefficients control position and the speed of the incline a N s |t is the average number of samples generated from sensor s in area a: a

a

N s |t = σ · Nsa |t + (1 − σ) · N s−1 |t I

I I

Nsa |t corresponds to the number of samples collected from sensor s in timeslot t in area a a N s−1 |t is its previous value σ is the exponential weighting coefficient


28 of 28

Smartphone Sensing Potential, sps I

Function of locally spent energy Es for sensing and reporting

I

Can be defined as the following sigmoid function: sps =

I I

1 1+e

− θζss ·(−Es +(1− θ2s ))

1 − θs /2 and ζs control position of the center and speed of the incline Es is the energy spent attributed to sensing (Esc ) and reporting (Esr ): Es = Esc + Esr

I I I

Esr depends on the employed communication technology (LTE or WiFi) c Esc = E s · Us Utilization context Us is defined as: ( 0, if the sensor s is used by another application Us = 1, otherwise


28 of 28

Threshold δ I

I

I

Depends on the Level of battery of the devices (B) and the Amount of reported data that devices have already contributed (D) Defined as: δ = f (δb , δd ) = (δb + δd )/2 Level of battery: δb = αλ·B I I

I

α can assume arbitrary real values between [0, 1] (α = 0.7) λ > 1 (λ = 10)

Amount of reported data:  1, if D ≥ Dmax δd = D log 1 + , otherwise Dmax X D= DsW + DsL s∈S


28 of 28

Evaluation Settings I

I I

FXOS8700CQ 3-axis linear accelerometer from Freescale Semiconductor BMP280 from Bosch for temperature and pressure Energy cost related to communication Esr depends on WiFi Z τtx Esr = PtxW dt 0

PtxW

= ρid + ρtx · τtx + γxg · λg


28 of 28

700

650

600

550

500

450

400

350

300

250

200

150

100

50

175 150 125 100 75 50 25 0

Density

Ostermalm Traces

Time Interval (sec) Andrea Capponi | GreenDays@Sophia 2017 | Evaluation of Energy-efficient Data Collection in MCS Systems with CrowdSenSim

28 of 28

Evaluation of Energy-efficient Data Collection in

Evaluation of Energy-efficient Data Collection in

Suggest Documents