Azeem Khan Kasthuri Jayarajah Dongsu Han Archan Misra Rajesh Balan Srinivasan Seshan Singapore Management University Carnegie Mellon University ID: 181004
Download Presentation The PPT/PDF document "CAMEO: A Middleware for Mobile Advertise..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
CAMEO: A Middleware for Mobile Advertisement Delivery
Azeem Khan†, Kasthuri Jayarajah*, Dongsu Han ‡, Archan Misra*, Rajesh Balan*, Srinivasan Seshan ‡* Singapore Management University‡ Carnegie Mellon University†Oriental Institute of ManagementSlide2
Motivations
Improving performance of mobile advertisements deliveryDecreasing bandwidth usageReducing energy consumption on mobileIntroduce monetization of advertisements by users and ISPsSlide3
Research Challenges
Reduce overheads in delivering adsProvide offline access to advertisement selectionFramework that enables dynamic negotiation in trading advertisements for connectivitySlide4
Background for addressing performance issues
Data collection process for advertisementsContexts: app, location, device type, OS, carriersPeriod: Every 1 minute, 2 weeks – 2 monthsProcedure: Scripts on computers in USA, Asia, EuropeObservationsTop 100 ads account for > 50% of views37% of ads are seen even after a day2/3 or more of ad content is redundant across ads (templated HTML)Country specific ads (overlap < 6%)
37% ads seen after a day
48
%
ads seen after
6
hoursSlide5
Background study for performance issues
Data collection procedure for usersWho: 20 participants on SMU campus for 1 monthProcedure: Custom LiveLabs app running as a monitoring service on Android 4.0+
Observations
On average, users switch between
WiFi
and 3G networks 2-4 times per day
Users are often connected to
WiFi
when charging phone
Users are on 3G network more than 50% of the time
Heavy
WiFi
usage
WiFi
connectedSlide6
Challenge #1:
Reduce overheadsHow?Pre-fetching and caching of advertisements.Why Both?pre-fetchingCAMEO exploits the fact that users are often on cheaper WiFi networks more than once a day!Advertisement contexts that matter such as location and app can be predictedcachingAds are repeatedSmall number of ads account for most ad views
Overheads per ad are avoidedSlide7
Caching and Pre-Fetching
CONTEXT PREDICTORAD MANAGER
APP #1
APP #2
CAMEO
AN
AN
AN = ADVERTISEMENT NETWORK
CACHE
More than 70% savings in
ad related bandwidth
is observed…Slide8
Challenge
# 2: Offline Access to AdsOnline selection of adsAN advertisement selection (ANAS)Bulk pre-fetch of ads, online ad selection by ANOffline selection of adsLocal advertisement selection (LAS)Bulk pre-fetch of ads, AN provides selection rulesBest effort advertisement selection (BEAS)Bulk pre-fetch of ads, statistical selection by CAMEOSlide9
Advertisement Selection
AD MANAGERAPP #1
APP #2
CAMEO
AN = ADVERTISEMENT NETWORK
CACHE
ACCOUNTING
&
VERIFICATION
RULESETSlide10
Energy gains by pre-fetching and caching
Base case measurement procedureScreen is lit (50% brightness on Samsung S3)No other app/services running except OS defaultWiFi of SMU campus, 3G on SingTel SingaporePre-fetching performed on cheaper network when phone is charging.ads fetched once every 45 seconds by custom appMonsoon monitoring device measures device power consumptionGains in LAS and BEAS for 1000 ad views for mostly offline apps99% savings in energy of radio useAnd nearly 92% savings in bandwidthSlide11
Challenge
# 3: Bartering ads for connectivityExample Scenario: A man walks into a airport where they charge $10 for connectivity. Would it be possible for him to get access in exchange for seeing advertisements from the airport’s network?Implications & AssumptionsForeground appsNegotiations are transparent to the userSlide12
Can we trade?
CAMEOISP2. NEGOTIATE
APP
OS
3. AD FETCH
1. BARTER?
4. AD(S)
5. BITS USEDSlide13
CAMEO architecture
CONTEXT PREDICTORAD MANAGER
ISP NEGOTIATOR
ACCOUNTING AND VERIFICATION
APP #1
APP #2
AN# 1 LIBRARY
AN #2 LIBRARY
CAMEOSlide14
Limitations of current CAMEO implementation
The user study is not representativeLong term context prediction may never be 100% accurateA small amount of space in memory will be occupied by the cache (approx. 2 MB for 1000 ad views)Accounting and verification need to be robust.These issues are currently under investigation.Slide15
Summary
#Challenge 1: Reduce overheadsPre-fetching and caching enable significant reduction in bandwidth and energy consumption #Challenge 2: Offline access of adsonline and offline modes of ad selection to preserve and enhance current economic models#Challenge 3: Framework for tradingInitial framework proposed and implemented*Thanks to Matt Welsh, the PC reviewers and my colleagues at SMU*Slide16
Questions?Slide17
Mobile Advertising Stakeholders
Bandwidth QuotaEnergy consumption
Signaling overhead
AN ≡ advertising networkSlide18
EMPIRICAL STUDY - Advertisements
Caching
could be very effective
Large amounts of
redundant information
Small percentage
of ads
dominate
viewsSlide19
EMPIRICAL STUDY - Users
Users are mostly on expensive networksUsers are price consciousSlide20
Design Goals
Lower cost of advertisement deliveryMinimize user involvementIncentivize developers to make applications consumer friendlyMinimal modifications to applications and mobile advertising networks.Slide21
CONTEXT PREDICTION
Algorithm to analyze and predict contextContext prediction accuracySlide22
CAN WE TRADE?
CAMEO
ISP
NEGOTIATOR
Negotiate
Accepted
Request Ad
Thanks for all the fish
Context Specific Ad
Bye
Accounting
APP
Register (1 ad, 10KB, TCP port 2894)
Display Ad
Success
Ad ready
Disconnect
ISP
G/W
ANDROID
OS
How
many
bytes?
Data transmission
10 KB, TCP 2894
IP A.B.C.D
Accounting
Close 2984
Completed