Augmenting mobile 3 g using wifi

Augmen'ng Mobile 3G Using WiFi

By: Aruna Balasubramanian, Ratul Mahajan, Arun Venkataramani

Presenter: Ke Huang

1Tuesday, April 12, 2011

Demand for mobile access growing www.totaltele.com

2

h+p://www.readwriteweb.com


Demand for mobile access growing www.totaltele.com

2

h+p://www.readwriteweb.com

900 million mobile broadband subscrip'ons today….

www.3gamericas.org


Mobile demand is projected to far

3



3

Current spectrum 409.5 MHz

Unallocated spectrum (including whitespaces)

230 MHz

Projected demand by 2016

800 MHz – 1000 MHz

www.rysavy.com



3



230 MHz


800 MHz – 1000 MHz

www.nyCmes.com

www.rysavy.com



“In light of the limited natural resource of spectrum, we have to look at the ways of conserving spectrum” -‐-‐ Mark Siegel (AT&T)

3



230 MHz


800 MHz – 1000 MHz

www.nyCmes.com

www.nyCmes.com

Reducing cellular spectrum u'liza'on is key!

www.rysavy.com


How can we reduce spectrum usage?

1. Behavioral

2. Economic

3. Technical

blogs.chron.com

4



1. Behavioral

2. Economic

3. Technical

blogs.chron.com

4

www.usatoday.com



1. Behavioral

2. Economic

3. Technical

blogs.chron.com

4

www.usatoday.com


Augmen'ng Mobile 3G using WiFi

Offload data to WiFi when possible

Focus on vehicular mobility

5


Offloading 3G data to WiFi

6



6



6



6



6



6


7

Related work on mul'ple interfaces

Improving performance using handoffs based on current condi'ons

Reducing power consump'on by switching across mul'ple interfaces


7

This work:1.How much 3G data can be offloaded to WiFi?2.How to offload without hur'ng applica'ons?

Related work on mul'ple interfaces

Improving performance using handoffs based on current condi'ons

Reducing power consump'on by switching across mul'ple interfaces


8

Contribu'ons

Measurement: Joint study of 3G and WiFi connec'vityAcross three ci'es: Amherst, Seagle, SFO

System: Wiffler, to offload 3G data to WiFi while respec'ng applica'on constraints Deployed on 20 vehicles


9

Measurement setup

Testbed: Vehicles with 3G and WiFi (802.11b) radiosAmherst: 20 buses + 1 car, Seagle: 1 car, SFO: 1 car

Soiware: Simultaneously probes 3G and WiFi for Availability, loss rate, throughput

Dura'on: 3000+ hours of data over 12+ days


Open WiFi availability low, but useful

10

Availability (%)

Availability = frac'on of 1-‐second intervals when at least one packet received



10

Availability (%)

86%




10

Availability (%)

86%

11%




10

Availability (%)

86%

11%


7%



10

Availability (%)

86%

11%


7%

3G+WiFi combinaCon be+er than sum pf parts


WiFi loss rate is higher

11

Cumulative fraction WiFi

3G

Loss rate = Frac'on of packets lost at 10 probes/sec



11


3G

28%




11


3G

28%

8%



WiFi (802.11b) throughput is lower

12

Cumulative fraction

Cumulative fraction

WiFi

3G

WiFi

3G

Upstream

Downstream

Throughput = Total data received per second



12

Cumulative fraction

Cumulative fraction

WiFi

3G

WiFi

3G

Upstream

Downstream




12

Cumulative fraction

Cumulative fraction

WiFi

3G

WiFi

3G

Upstream

Downstream

0.35




12

Cumulative fraction

Cumulative fraction

WiFi

3G

WiFi

3G

Upstream

Downstream

0.35 0.72




12

Cumulative fraction

Cumulative fraction

WiFi

3G

WiFi

3G

Upstream

Downstream

0.35 0.72



13

Implica'ons of measurement study

Strawman augmenta'on: Use WiFi when availableCan offload only ~11% of the 'meCan hurt applica'ons because of WiFi’s higher loss rate and lower throughput


14

Key ideas in Wiffler

Increase savings for delay-‐tolerant applica'ons

Problem: Using WiFi only when available saves ligle 3G usage

Solu'on: Exploit delay-‐tolerance to wait to offload to WiFi when availability predicted

Reduce damage for delay-‐sensi've applica'ons

Problem: Using WiFi whenever available can hurt applica'on quality

Solu'on: Fast switch to 3G when WiFi delays exceed threshold


Predic'on-‐based offloading

D = Delay-‐tolerance threshold (seconds)S = Data remaining to be sent (bytes)

Each second,1. If (WiFi available), send data on WiFi 2. Else if (W(D) < S), send data on 3G3. Else wait for WiFi.

15


Predic'on-‐based offloading

D = Delay-‐tolerance threshold (seconds)S = Data remaining to be sent (bytes)

Each second,1. If (WiFi available), send data on WiFi 2. Else if (W(D) < S), send data on 3G3. Else wait for WiFi.

15

Predicted WiFi transfer size in next D seconds


16

Predic'ng WiFi capacity

History-‐based predic'on of # of APs using last few AP encounters WiFi capacity = (expected #APs) x (capacity per AP)

Simple predictor yields low error both in Amherst and Seagle


16

Negligible benefits with more sophis'cated predic'on, eg future loca'on predic'on + AP loca'on database

Predic'ng WiFi capacity

History-‐based predic'on of # of APs using last few AP encounters WiFi capacity = (expected #APs) x (capacity per AP)

Simple predictor yields low error both in Amherst and Seagle


17

Fast switching to 3G

Problem:WiFi losses bursty => high retransmission delay

Approach:If no WiFi link-‐layer ACK within 50ms, switch to 3GElse, con'nue sending on WiFi


Wiffler implementa'on

18

Wiffler proxy

§ Predic'on-‐based offloading upstream + downstream§ Fast switching only upstream

Ø Implemented using signal-‐upon-‐ACK in driver


19

Evalua'on Roadmap

Predic'on-‐based offloadingDeployment on 20 DieselNet buses in 150 sq. mi region around Amherst

Trace-‐driven evalua'on using throughput data

Fast switchingDeployment on 1 car in Amherst town centerTrace-‐driven evalua'on using measured loss/delay trace using VoIP-‐like probe traffic


Deployment resultsData offloaded to WiFi

Wiffler’s predic'on-‐based offloading 30%WiFi when available 10%

20

File transfer size: 5MB; Delay tolerance: 60 secs; Inter-‐transfer gap: random with mean 100 secs


Deployment resultsData offloaded to WiFi

Wiffler’s predic'on-‐based offloading 30%WiFi when available 10%

20

% Cme good voice quality Wiffler’s fast switching 68%

WiFi when available (no switching) 42%

File transfer size: 5MB; Delay tolerance: 60 secs; Inter-‐transfer gap: random with mean 100 secs

VoIP-‐like traffic: 20-‐byte packet every 20 ms


21

Trace-‐driven evalua'on

Parameters variedWorkload, AP density, delay-‐tolerance, switching threshold

Strategies compared to predic'on-‐based offloading:WiFi when availableAdapted-‐Breadcrumbs: Future loca'on predic'on + AP loca'on database

Oracle (Imprac'cal): Perfect predic'on w/ future knowledge


Wiffler increases data offloaded to WiFi

22

Workload: Web traces obtained from commuters



22


14%WiFi when available yields ligle savings



22


42%

14%

Wiffler close to Oracle



22


42%

14%

Wiffler close to OracleSophis'cated predic'on yields negligible benefit



22


Wiffler increases delay by 10 seconds over Oracle.

42%

14%

Wiffler close to OracleSophis'cated predic'on yields negligible benefit


Even more savings in urban centers

23


Fast switching improves quality of delay-‐sensi've applica'ons

24



24

58%



24

40%58%



24

40%58%

73%



24

40%58%

73%

30% data offloaded to WiFi with 40ms switching threshold


25

Future work

Reduce energy to search for usable WiFi

Improve performance/usage by predic'ng user accesses to prefetch over WiFi

Incorporate evolving metrics of cost for 3G and WiFi usage


26

Summary

Augmen'ng 3G with WiFi can reduce pressure on cellular spectrum

Measurement in 3 ci'es confirms WiFi availability and performance poorer, but poten'ally useful

Wiffler: Predic'on-‐based offloading and fast switching to offload without hur'ng applica'ons


26

Summary

Augmen'ng 3G with WiFi can reduce pressure on cellular spectrum

Measurement in 3 ci'es confirms WiFi availability and performance poorer, but poten'ally useful

Wiffler: Predic'on-‐based offloading and fast switching to offload without hur'ng applica'ons

Questions?


Thank you!


Technology

Augmenting mobile 3 g using wifi