Upload
others
View
4
Download
0
Embed Size (px)
Citation preview
IEEE ICC 2015
Network-Assisted Offloading for MobileCloud Applications
Claudio Fiandrino
Dzmitry Kliazovich
Pascal Bouvry
University of Luxembourg
Albert Y. Zomaya University of Sydney
June 9, 2015
Motivation
I 4.4 billion people will use mobile cloud applications by 2017I $ 45 billion marketI Mobile cloud applications: 90% of all mobile data traffic by 2019
2014 2015 2016 2017 2018 20190
50%
100%19% 17% 15% 14% 12% 10%
81% 83% 85% 86% 88% 90%
Non-CloudCloud
Source: Cisco Visual Networking Index: Global Mobile Data Traffic Forecast Update, 2014-2019
Claudio Fiandrino | IEEE ICC 2015 | Network-Assisted Offloading for Mobile Cloud Applications 1 of 11
Mobile data traffic
1 EB
30 EB
24.3 EB
Global Internet2000
Mobile Networks2014
Mobile Networks2019 (per month)
Source: Cisco Visual Networking Index: Global Mobile Data Traffic Forecast Update, 2014-2019
Claudio Fiandrino | IEEE ICC 2015 | Network-Assisted Offloading for Mobile Cloud Applications 2 of 11
Techniques
Increasing capacity
I Deploying more base stations(Micro, Pico, Femto)
3 Improve coverage
7 Cost of installation and mainte-nance
Traffic Offloading
I Offload traffic to other networks(WiFi, opportunistic)
3 Use of existing infrastructure
7 Use of different technologiesthan cellular
Claudio Fiandrino | IEEE ICC 2015 | Network-Assisted Offloading for Mobile Cloud Applications 3 of 11
Traffic Offloading
Bear in mind:I Performance of applications
I Bandwidth, latencyI User profiles
I Mobility, behaviour
Claudio Fiandrino | IEEE ICC 2015 | Network-Assisted Offloading for Mobile Cloud Applications 4 of 11
The scenario
I Network-assisted offloading over WiFi networks
Mobile Operator Network
CloudInternetP-GWS-GW
MCOHMME
HSS PCRFeNodeB
IP Network
WiFi APUEs
Mobile Cloud Offloading Helper
I Software module with high computing capabilities
I Offloading balancing application requirementsuser behaviour
and network resources
Claudio Fiandrino | IEEE ICC 2015 | Network-Assisted Offloading for Mobile Cloud Applications 5 of 11
The scenario
I Network-assisted offloading over WiFi networks
Mobile Operator Network
CloudInternetP-GWS-GW
MCOHMME
HSS PCRFeNodeB
IP Network
WiFi APUEs
Mobile Cloud Offloading Helper
I Software module with high computing capabilities
I Offloading balancing application requirementsuser behaviour
and network resources
Claudio Fiandrino | IEEE ICC 2015 | Network-Assisted Offloading for Mobile Cloud Applications 5 of 11
Offloading decision
UserBehavior
Mobile CloudOffloading
Helper (MCOH)
ApplicationRequirements
Availabilityof NetworkResources
MobilityData Plan
Channel QualityRate
Latency
Channel QualityRate
Latency
Claudio Fiandrino | IEEE ICC 2015 | Network-Assisted Offloading for Mobile Cloud Applications 6 of 11
Evaluation
I NS-3 SimulationsI Scenario:
I 4 users equipped with both LTE and WiFiI Average mobility speed: [1.4,10] m/sI Available data plan: [0,1] GbI VOIP application (64 Kbit/s)
Avg. Speed: 7.54 m/sAv. Data: 0.48 GB
Avg. Speed: 4.23 m/sAv. Data: 0.87 GB
Avg. Speed: 2.27 m/sAv. Data: 0.78 GB
Avg. Speed: 4.78 m/sAv. Data: 0.71 GB
Claudio Fiandrino | IEEE ICC 2015 | Network-Assisted Offloading for Mobile Cloud Applications 7 of 11
Amount of offloaded traffic
Avg. Speed: 7.54 m/sAv. Data: 0.48 GB
Avg. Speed: 4.23 m/sAv. Data: 0.87 GB
Avg. Speed: 2.27 m/sAv. Data: 0.78 GB
Avg. Speed: 4.78 m/sAv. Data: 0.71 GB
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
0.5
1
1.5
2
2.5
3
3.5
·104
1− γ
Offl
oade
dtr
affic
[Byt
es]
Claudio Fiandrino | IEEE ICC 2015 | Network-Assisted Offloading for Mobile Cloud Applications 8 of 11
Mobility
0 0.2 0.4 0.6 0.8 1
0
0.2
0.4
0.6
0.8
1
Mobility factor
Traf
ficra
tio
Theoretical Experimental
I Velocity is a highly dynamic parameter
Claudio Fiandrino | IEEE ICC 2015 | Network-Assisted Offloading for Mobile Cloud Applications 9 of 11
Data plan
0 0.2 0.4 0.6 0.8 1
0
0.2
0.4
0.6
0.8
1
Data plan factor
Traf
ficra
tio
Theoretical Experimental
I Amount of available data is not dynamic
Claudio Fiandrino | IEEE ICC 2015 | Network-Assisted Offloading for Mobile Cloud Applications 10 of 11
Conclusion
Summary
I Software module to take offloading decisionsI Use information already present in operator networks
Take home message
I Network- and application-awareness crucial for offloadingdecisions
Claudio Fiandrino | IEEE ICC 2015 | Network-Assisted Offloading for Mobile Cloud Applications 11 of 11