1. Augmenting Mobile 3G Using WiFiby
Aruna Balasubramanian
Ratul Mahajan Arun Venkataramani
University of Massachusetts Microsoft Research
Presented by
Ashok Kumar J
CS 752/852 - Wireless and Mobile Networking

2. Demand for mobile access growing
Revenue from mobile data will overtake fixed voice in the US in 2011, according to Pyramid Research.
All of this is understandable given the massive adoption of mobile devices such as smartphones.
900 million mobile broadband subscriptions today….

3. Mobile demand is projected to far exceed capacity
Current spectrum
409.5 MHz
Unallocated spectrum (including whitespaces)
230 MHz
Projected demand by 2016
800 MHz – 1000 MHz
Demand can never be met given the drawn out process of reclaiming spectrum.
Carriers understand this and are already looking for other mechanisms.
4G reaches shannon limit with snr. only can do spatial reuse -- reducing radius of a cell,
All projections are based on current networks. But actual demand may be even higher.
“In light of the limited natural resource of spectrum, we have to look at the ways of conserving spectrum” -- Mark Siegel (AT&T)
Reducing cellular spectrum utilization is key!

4. Demand projected to outstrip capacity
Per person estimate by Peter Rysavy, based on allocated and allocated but unused capacity
In our smartphone measurements, some users are already consuming about 0.5GB a day.

5. How can we reduce spectrum usage?
blogs.chron.com
1. Behavioral
2. Economic
3. Technical
All three are probably needed.
Our focus is on reducing usage.

6. Augmenting Mobile 3G using WiFi
Offload data to WiFi when possible Focus on vehicular mobility
We are surrounded by WiFi and most devices have multiple interfaces.
Inspired by home
Particularly challenging case and increasing demand because of navigation-based applications. Offloading is less of a challenge when you are stationary – you can either do it or not – we wanted to see how much data can you offload when you are out and about.
Agnostic to business model

7. Offloading 3G data to WiFi

8. Related work on multiple interfaces
Improving performance using handoffs based on current conditions
Reducing power consumption by switching across multiple interfaces
This work:
How much 3G data can be offloaded to WiFi?
How to offload without hurting applications?

9. Contributions Measurement: Joint study of 3G and WiFi connectivity
Across three cities: Amherst, Seattle, SFO
System: Wiffler, to offload 3G data to WiFi while respecting application constraints
Deployed on 20 vehicles
WiFi APs have short range and are typically not deployed to cover roads. Quality might be very poor.
The study shows that WiFi availability is low and its quality is poor.
A simplistic design …………..
With this in mind, we design Wiffler

10. Measurement setup Testbed: Vehicles with 3G and WiFi (802.11b) radios
Amherst: 20 buses + 1 car, Seattle: 1 car, SFO: 1 car
Software: Simultaneously probes 3G and WiFi for
Availability, loss rate, throughput
Duration: hours of data over 12+ days

11. Open WiFi availability low, but useful
Availability = fraction of 1-second intervals when at least one packet received
86%
Availability (%)
3G+WiFi combination better than sum pf parts
11% 7%

12. WiFi loss rate is higher
Loss rate = Fraction of packets lost at 10 probes/sec
Cumulative fraction
28%
WiFi 8% 3G
Wi-Fi loses are bursty.

13. WiFi (802.11b) throughput is lower
Throughput = Total data received per second
Cumulative fraction
WiFi
Upstream 3G
0.35
0.72
Cumulative fraction
WiFi
Downstream 3G

14. Interesting observations…
Availability of 3G in Peak hours is less compared to its availability in non-peak hours.
Availability of Wi-Fi in Peak hours is more compared to its availability in off-Peak hours.
Unavailability of 3G is 11% but when combined with Wi-Fi, total unavailability reduced to 5%.

15. Interesting observations…
In 47% of the Locations, Data sent over Wi-Fi is insignificant compared to 3G.
In remaining 53% of locations, at least 20% of the 3G data could be shifted to Wi-Fi.
In 9% of the locations, equal or more data sent over Wi-Fi compared 3G. So in these location entire traffic could be offloaded to Wi-Fi.

16. Implications of measurement study
Strawman augmentation: Use Wi-Fi when available
Can offload only ~11% of the time
Can hurt applications performance because of Wi-Fi's higher loss rate and lower throughput
Example: Applications sensitive to losses like VoIP and Video Conferencing will face degraded application quality while transmitting over Wi-Fi but Application like , sending a file wouldn’t.

17. Key ideas in Wiffler Fast Switching to 3G: Leveraging Delay Tolerance:
Reduce damage for delay-sensitive applications
Problem: Using WiFi whenever available can hurt application quality
Solution: Fast switch to 3G when WiFi delays exceed threshold
Leveraging Delay Tolerance:
Increase savings for delay-tolerant applications
Problem: Using WiFi only when available saves little 3G usage
Solution: Exploit delay-tolerance to wait to offload to WiFi when availability predicted
Not worrying about energy for now.
These techniques are very simple and yet close to optimal for what they do. In the beginning, we played with some more complicated versions but found them to not worth it.

18. Prediction-based offloading
D = Delay-tolerance threshold (seconds)
S = Data remaining to be sent (bytes)
Each second,
If (WiFi available), send data on WiFi
Else if (W(D) < S), send data on 3G
Else wait for Wi-Fi.
Choosing a delay threshold involves a trade off between better application performance and 3G load.
Predicted WiFi transfer size in next D seconds

19. Predicting WiFi capacity
History-based prediction of # of Aps encounter until a future time using average inter-meeting time of the past encounters. (T/I encounter in next Tsecs where I is avg)
Similarly average throughput is estimated based on the throughput observed by each vehicle at each AP encounter.
WiFi capacity = (expected #APs) x (capacity per AP)
Negligible benefits with more sophisticated prediction, eg future location prediction + AP location database

20. Error in predicting # of APs
Relative error
N=4
N=8

21. Predicting WiFi capacity
Accuracy is Low when predicting with only one previous encounter.
Predicting error is close to 20% even for predicting Ap encounter untill small future time interval of 20 secs.
When prediction is based on the previous 4 or 8 AP encounters, the predictions error is less than 5% up to a future prediction time-interval of 50 seconds.
The prediction error increases to 20% for prediction time interval of 100sec.

22. Fast switching to 3G Problem: Approach:
WiFi losses bursty => high retransmission delay
Approach:
If no WiFi link-layer ACK within 50ms, switch to 3G
Else, continue sending on WiFi
Wi-Fi NIC frequently takes a long time to complete retransmission attempts. Madwifi which used in test beds retries packets 11 times, which even if we ignore medium access delays takes more than 120 milliseconds with default b specification.
So fast switching mechanism send the packet on 3G if the Wi-Fi link-layer fails to deliver the packet within a delay threshold.
A different way of bonding multiple interfaces

23. Adopting Wiffler
Wiffler needs to know the delay tolerance threshold and the QoS requirements of each application that uses the network.
Wiffler requires proxy support, both to impliment fast switching and the prediction-based offloading. Proxy will fecilitate packet reception from multiple IP address (ie from the 3G and the Wifi interfaces) and allow switching between interfaces.
Experiments are based open WiFi APs, Wiffler can be deployed used with other APs as well.
A different way of bonding multiple interfaces

24. Wiffler implementation
Wiffler proxy
Destination server also acts as proxy to manage data coming from different IP address that the client acquires as it moves.
Prediction-based offloading upstream + downstream
Delay threshold to 50 ms.
A proxy is coming for a host of other reasons as well.

25. Wiffler implementation
Wiffler proxy
Added a signal mechanism in the mobile node’s driver that signals the application when the wireless card receives a link layer acknowledgement.
Fast switching only upstream
Fast switching in downstream direction is challenging because it needs either support from the APs or detailed information at the proxy on current Wi-Fi conditions(conveying the same is time consuming).
A proxy is coming for a host of other reasons as well.

26. Evaluation Roadmap Prediction-based offloading Fast switching
Deployment on 20 DieselNet buses in 150 sq. mi region around Amherst
Trace-driven evaluation using throughput data
Fast switching
Deployment on 1 car in Amherst town center
Trace-driven evaluation using measured loss/delay trace using VoIP-like probe traffic

27. % time good voice quality
Deployment results
Data offloaded to WiFi
Wiffler’s prediction-based offloading
30%
WiFi when available
12%
File transfer size: 5MB; Delay tolerance: 60 secs; Data generation gap: random with mean 100 secs
% time good voice quality
Wiffler’s fast switching
68%
WiFi when available (no switching)
42%
VoIP-like traffic: 20-byte packet every 20 ms

28. Trace-driven evaluation
Parameters varied
Workload, AP density, delay-tolerance, switching threshold
Strategies compared to prediction-based offloading:
WiFi when available
Adapted-Breadcrumbs: Future location prediction + AP location database
Oracle (Impractical): Perfect prediction w/ future knowledge

29. Wiffler increases data offloaded to WiFi
Workload: Web traces obtained from commuters
Wiffler close to Oracle
42%
Sophisticated prediction yields negligible benefit
14%
WiFi when available yields little savings
Wiffler increases delay by 10 seconds over Oracle.

30. Even more savings in urban centers
Completion Time
Even more savings in urban centers

31. Fast switching improves quality of delay-sensitive applications
73%
58%
40%
30% data offloaded to WiFi with 40ms switching threshold

32. Future work Reduce energy to search for usable WiFi
Improve performance/usage by predicting user accesses to prefetch over WiFi
Incorporate evolving metrics of cost for 3G and WiFi usage

33. Summary
Augmenting 3G with WiFi can reduce pressure on cellular spectrum
Measurement in 3 cities confirms WiFi availability and performance poorer, but potentially useful
Wiffler: Prediction-based offloading and fast switching to offload without hurting applications

34. Questions?