Upload
theus-hossmann
View
894
Download
0
Tags:
Embed Size (px)
Citation preview
Christian RohnerPer Gunningberg
Uppsala Universitet, Sweden
Twitter in Disaster Mode:Opportunistic Communication and
Distribution of Sensor Data in Emergencies
Theus HossmannFranck Legendre
Paolo CartaETH Zurich, Switzerland
Source: XKCD (http://xkcd.com/723/)
Source: Twitter Blog (http://blog.twitter.com/2011/06/global-pulse.html)
Network Outage in Japan
Operator # inoperative BS
NTT DoCoMo 6720
KDDI 3800
Softbank 3786
Your Smart Phone, the Emergency Kit
• Temporary GSM network• Wireless mesh network• Satellite communication
• Opportunistic Communication• DTN2• Haggle• PodNet
Deployment, configuration, etc.• Requires experts• > 1-2 days
✗• No expert skills required• Instantly ready✓
Goal: Enable disaster victims to tweet instantaneouslyGoal: Enable disaster victims to tweet instantaneously
Twimight
• Simple yet flexible• Wide spread (200M users)
• Simple yet flexible• Wide spread (200M users)
• Wide spread • Developer friendly• Wide spread • Developer friendly
• Disaster Mode (user enabled with a simple settings check-box)
✓ Opportunistic Communication
✓Spreading of sensor data
• Disaster Mode (user enabled with a simple settings check-box)
✓ Opportunistic Communication
✓Spreading of sensor data
• Open source (Google Code)
• Open source (Google Code)
Opportunistic Spreading of Tweets
• Bluetooth communication• Requires pairing• Scanning (2min ± 20sec)
• Power saving heuristic• Reduced scanning interval at battery levels < 50%• No more scanning at levels below 30%
• Epidemic spreading (flooding)• Small data volumes• FIFO buffer
• Publish tweets once connectivity is [email protected]
Tweeting of Sensor Data
• Twimight plugin: Accessing disaster tweet DB• User activity detection
• Monitoring acceleration sensor• Classify: {Still | walking | running}• Nearest Neighbor classification
• Environment detection• Monitoring microphone• Classify: {Silence | Speech | Music}• Decision tree classification
• Posting tweet on activity change
Das et al. “Detecting User Activities using the Accelerometer on Android Smartphones“
Das et al. “Detecting User Activities using the Accelerometer on Android Smartphones“
Lu et al. “SoundSense: Scalable Sound Sensing for People-CentricApplications on Mobile Phones“
Lu et al. “SoundSense: Scalable Sound Sensing for People-CentricApplications on Mobile Phones“
What’s next?
• Security (in progress)• Asymmetric crypto for signing (tweets) and encrypting (DM)• “Pragmatic” hybrid approach: Key distribution before
disaster• Twitter Disaster Server
• Smart tweet delivery• Contact Graph based routing for Direct Messages• Interest matching for tweets
• In parallel• Improvements in User Interface• Advanced Plugin System: Browse and install plugins within
Twimight• Plugins, plugins, plugins…
Get Involved
• Twimight: http://code.google.com/p/twimight• Sensor:
http://code.google.com/p/android-sensor-twitteth/
Video
Thank You For Installing & Using Twimight