The Vobile Cloud : Cars are Better for Mobile Clouds and Group Connect

.

The Vobile Cloud

M.Zhanikeev -- [email protected] The Vobile Cloud : Cars are Better for Mobile Clouds and Group Connect -- bit.do/marat141218 -- 2/23...

2/23

.

The Mobile Cloud

WiFi

3G

WiFi

3G 3G

•GroupConnectis virtualizationbetween localand remoteconnectivity

• WirelessUniversityCampus 02

• streaming tomassivecrowds 03

02 myself+0 "Opportunistic Multiconnect with P2P WiFi and Cellular Providers" Mobile Computing... 4G and Beyond, CRC (2015)

03 myself+0 "Video Streaming overs Wireless Channels -- 4G and Beyond" Mobile Computing... 4G and Beyond, CRC (2015)


3/23

.

The Vobile Cloud

WiFi

3G

WiFi

3G

3G

• NOTE: carshave infinitebatteries 01

• mobility isdifferent

• otherwise thesame

01 myself+0 "Vehicular Clouds based on GroupConnect and Self Organization" Cognitive Vehicular Networks, CRC (2015)


4/23

.

Vehicular Clouds : Literature

• V2C is the main mode 10

• cloud for vehicles = information support 11◦ messages from CAN bus are uploaded and processed in the cloud 12

• closest to this report: various *aaS services based on cars 09

10 R.Yu+4 "Toward Cloud-based Vehicular Networks with Efficient Resource Management" IEEE Networks (2013)

11 M.Gerla+3 "Internet of Vehicles: From Intelligent Grid to Autonomous Cars and Vehicular Clouds" W.Sens.Net. Mag. (2014)

12 Y.Xu+1 "A Cloud Based Information Integration Platform for Smart Cars" 2nd ... Urban Computing and Smart Grid (2011)

09 M.Whaiduzzaman+3 "A Survey on Vehicular Cloud Computing" Journal of Network and Computer Applications (2014)


5/23

.

What's Missing

1. V2C singletons rather than VANETs2. no connection between VANETs and clouds3. no connection 5G which has D2D and M2M


6/23

.

Vobile: What's Proposed

1. GroupConnect is the main mode (better than VANETs) 01 02

2. the Mobile Cloud approach: local + remote + sync3. many applications: cloud multisource 04, client-side aggregation 05

4. large-scale realistic models

01 myself+0 "Vehicular Clouds based on GroupConnect and Self Organization" Cognitive Vehicular Networks, CRC (2015)

02 myself+0 "Opportunistic Multiconnect with P2P WiFi and Cellular Providers" Mobile Computing... 4G and Beyond, CRC (2015)

04 myself+0 "Multi-Source Stream Aggregation in the Cloud" Advanced Content Delivery..., Wiley (2014)

05 myself+0 "Experiments with application throughput in a browser with full HTML5 support" IEICE ComEx (2013)


7/23

.

Vobile: Taxonomy and UsecasesCars

Rarely Meet

Meeting Likely

Travel Together

Park Together

DTN MANETs

Traditional SingleConnect

Sensor Cloud

Storage Cloud

Perfectly Vobile


8/23

.

Traces: Crawdad


9/23

.

Traces: Rome Taxies (crawdad)

• the closest thing, real Rome taxitraces 13

• about 300 taxies for 1 month• plot: 10 randomly selected routes

• plot: raw routes

13 A.Saha+1 "Modeling mobility for vehicular ad-hoc networks" 1st ACM VANET (2004)


10/23

.

Traces: Rome Taxies : Hot Meets

• GroupConnect depends on wirelessmeetings

• even with 300 cars, meetings arecommon

• why: popular routes, popular locations,bypasses, etc.

• hot meets: each place wheremeetings are common is a hotspot


11/23

.

Traces: Rome Taxies : Hot Stills

• hot stills: places where cars spendat least an hour together

• relatively rare

• why: 300 is too few cars


12/23

.

Rome Trace is not Enough

• possibly realistic hot meets, but not hot stills• cannot create a model where a larger fleet (common population) implementsVobile Cloud

• currently available traces and models are not sufficient


13/23

.

Related Modeling


14/23

.

Related : EV Infrastructure

• a model of a city-scale EVinfrastructure 06

• EV battery replacementand V2H use

• target: social utility


15/23

.

Related : Maps2Graphs• socially distributed generation oflarge GIS datasets (graphs)

• plot: a graph connecting 375 FamilyMarts in Kyushu

07 myself+0 "Maps2Graphs: A Socially Scalable Method for Generating High-Quality GIS Datasets..." ITS研 (2013)


16/23

.

Again, NOT Enough

• previous grain: EVs, cars

• needed grain: cloud services on top of groups of cars• .... much much much higher complexity


17/23

.

Modeling: the Vision


18/23

.

Traces and Synthesis

Traces

Travel Time

Density Centers

Mobility Traces

Synthesis

Context

Voronoi Plots

Hotspot Model


19/23

.

Traces : Travel Time

• same as in 06

• can be used to guess hot stills, butnot hot meets

• depends on density model --Maps2Graphs

06 myself+0 "On How Smart Cities Can Improve Social Utility of Their Citizens' Commutes" IPSJ JIP (2014)

07 myself+0 "Maps2Graphs: A Socially Scalable Method for Generating High-Quality GIS Datasets..." ITS研 (2013)


20/23

.

Hotspot Distribution

0 10 20 30 40 50List of traffic sources

0

0.4

0.8

1.2

1.6

2

2.4

2.8

log(

traf

fic v

olum

e)

0 10 20 30 40 50List of traffic sources

0

0.4

0.8

1.2

1.6

2

2.4

2.8

log(

traf

fic v

olum

e)

• 4 sets:1. normal versus hot2. normally hot

versus FlashCrowds

• applies to severalparts of the model

1. tenant vs1-story houses,

2. parking spaces

• directly relates to hotstills -- cloud storageservice


21/23

.

Wrapup

• working on the above large-scale model now• several components are ready and/or publically available

• will report on progress in future publications


22/23

.

That’s all, thank you ...


23/23

.

[01] myself+0 (2015)Vehicular Clouds based on GroupConnect and Self OrganizationCognitive Vehicular Networks, CRC

[02] myself+0 (2015)Opportunistic Multiconnect with P2P WiFi and Cellular ProvidersMobile Computing... 4G and Beyond, CRC

[03] myself+0 (2015)Video Streaming overs Wireless Channels -- 4G and BeyondMobile Computing... 4G and Beyond, CRC

[04] myself+0 (2014)Multi-Source Stream Aggregation in the CloudAdvanced Content Delivery..., Wiley

[05] myself+0 (2013)Experiments with application throughput in a browser with full HTML5 supportIEICE ComEx


23/23

.

[06] myself+0 (2014)On How Smart Cities Can Improve Social Utility of Their Citizens' CommutesIPSJ JIP

[07] myself+0 (2013)Maps2Graphs: A Socially Scalable Method for Generating High-Quality GISDatasets...ITS研

[08] myself+1 (2012)Popularity-Based Modeling of Flash Events in Synthetic Packet TracesCQ研

[09] M.Whaiduzzaman+3 (2014)A Survey on Vehicular Cloud ComputingJournal of Network and Computer Applications

[10] R.Yu+4 (2013)Toward Cloud-based Vehicular Networks with Efficient Resource ManagementIEEE Networks


23/23

.

[11] M.Gerla+3 (2014)Internet of Vehicles: From Intelligent Grid to Autonomous Cars and VehicularCloudsW.Sens.Net. Mag.

[12] Y.Xu+1 (2011)A Cloud Based Information Integration Platform for Smart Cars2nd ... Urban Computing and Smart Grid

[13] A.Saha+1 (2004)Modeling mobility for vehicular ad-hoc networks1st ACM VANET

[14] R.Amici+5 (2014)Performance assessment of an epidemic protocol in VANET using real tracesMoWNeT

[15] N.Fernando+3 (2013)Mobile cloud computing: A surveyElsevier ... Future Generation Computer Systems


23/23

.

[16] M.Satyanarayanan+3 (2009)The case for VM-based cloudlets in mobile computingIEEE Pervasive Computing

[17] G.Huerta-Canepa+1 (2010)A virtual cloud computing provider for mobile devices1st ACM MCS

[18] L.Deboosere+4 (2011)Grid design for mobile thin client computingFuture Generation Computer Systems

[19] A.Konak+2 (2013)A flocking-based approach to maintain connectivity in mobile wireless ad hocnetworksApplied Soft Computing


23/23

The Vobile Cloud : Cars are Better for Mobile Clouds and Group Connect

Technology

mobile clouds

mobile cloud approach

cloud multisource

g wifi

mobile cloudwifi

vobile cloudwifi

vehicular clouds w

vobile cloudm