PRESENTATION BY: SCOTT COREY REPORT BY: A. KHAN, V. SUBBARAJU, A. MISRA, S. SESHAN Mitigating the...
-
Upload
barry-hubbard -
Category
Documents
-
view
212 -
download
0
Transcript of PRESENTATION BY: SCOTT COREY REPORT BY: A. KHAN, V. SUBBARAJU, A. MISRA, S. SESHAN Mitigating the...
PRESENTATION BY: SCOTT COREYREPORT BY: A . KHAN, V. SUBBARAJU, A .
MISRA, S. SESHAN
Mitigating the True Cost of Advertisement- Supported“Free” Mobile Applications
“Free” Applications
Consumers have grown accustomed to “free” applications supported by advertisements
Free versions of Angry Birds earns $1 million+ a month in advertisements
Total free apps downloaded 81%
Free apps in top grossing list 51%
Free apps in iTunes Store 63%
Free apps in Android Market 37%
Free apps in Blackberry World 82%
Data Traffic
Consumers data usage is growing exponentially
Many major providers now have metered data plans
Most devices have alternative connections that are free Wi-Fi Femtocell
CAMEO
Context-Driven Advertisement Modulator and Optimizer
The current “free” application model is unsustainable
Inexpensive and unmetered bandwidth is readily available
Advertisements can be prefetched during these opportune times
The advertisement served is context-dependent
Characterization of Mobile Advertising Traffic
Began with experiment using “free” apps on a Nexus One
Applications were tested with 15 minutes of runtime (30 minutes for “Friends” games)
Network traffic was recorded to identify and compare usage from advertisements or the application itself
Results
Non-negligible amount of traffic from each application
Usage varied based on quantity of advertisements
Math
Fruit Ninja – 5.61 Kbps on averagePlayed for 30 minutes per dayOn Verizon Wireless
Current data plan: $30 for 2GB per monthThe advertisements alone generate ~40MB of
traffic, roughly 56 cents of dataPaid version is 99 cents
Trending Upwards
Advertisements are getting larger, more interactive
HTML5 based ads are typically 5-20 times larger
Multiple applications can be running at one time
“Free” applications are not free!
Key Characteristics of Mobile Advertising
Combination of users profile and recent environmental context
Context
Network phone is connected to Ads for other carriers
Location Ads for city events
Applications used Ads for similar applications and same developers
Model of the phone Ads for newer phones
Mechanics of Advertisement Delivery
Advertisements were overlaid overtop applications
Most applications used multiple advertisement networks
Multiple connections were necessary Contact network Get XHTML or Javascript code Fetch actual advertisement
Back to CAMEO
Must predict likely future context of the user/device
Intelligently utilizing available alternative networks
Serving advertisements from a local cache
Compatible with existing deployments
Must reduce cost of current advertising methods
Interception Switch to CAMEO from connecting to outside
networksAdvertisement Generation
Must have cached appropriate advertisements in advance
Combination of cache manager, ad-prefetcher, and context predictor
Cannot distinguish advertising trafficCannot verify if correct advertisements were
provided
Clean-Slate Design
Application Modifications Can query CAMEO components directly Applications verify advertisements are received and
appropriateInfrastructure Modifications
Data prefetching can be done in a network-friendly fashon
Networks can provide additional information to enhance context
Opens up advanced possibilities of advertising