Team 7: Sunoco Hotdog Eye Home MidSemester Presentation

1
Team 7 Sunoco HotdogEye HomeMid-Semester
Presentation

• Hong, Keetaek
• Baek, Jinwook
• Lee, David

2
Status Update

• Concept
• Home monitoring system that notifies via SMS
• Status Update of Implementation
• Done with basic implementation
• Now working on calibration and fault-tolerance
• Status Update of Testing and Experimentation
• Tested sensor sensitivity
• Tested normal operation
• Tested some fault-injected test scenarios

3
Experimentation Plan

• Test Setup
• 2 sensor Tmotes, 1 server Tmote, 1 PC, 1
  SMS-enabled mobile phone, Internet service, SMS
  gateway
• Metrics
• Delivery time End time Start time
• Start time Sensor detects smoke
• End time Cellphone receives SMS
• Binary
• 0 Not detected (SMS doesnt arrive within 2
  min)?
• 1 Detected (SMS arrives within 2 min)?
• Workload
• Up to two problems detected simultaneously

4
Experimentation Plan

• Parameters
• Location (humidity, lighting, pressure, etc.),
  size of fire, distance, radio reception, Internet
  connection
• Test run
• Perform in same location
• Simulate same fire size
• Keep Internet connection connected
• Keep number and nature of problems constant
• Vary distance from sensor the problem occurs
• Experiment
• Number of test runs 50 simulations
• Average Number of Messages 1 per simulation
• Min time 0 seconds, Max time 2 minutes

5
Initial Data

• Initial Data for the Smoke Sensor
• Due to poor sensor sensitivity, we allowed a very
  large range of analog values to be problems
• Distance (d) was 10 cm from sensor
• System successfully notified user under normal
  operation - 15 out of 15 times
• System successfully notified user upon
  reconnection after being disconnected from the
  Internet - 15 out of 15 times

6
Initial Data

• Distance Detected
• 0 Success
• 5 Success
• 10 Success
• 20 Success
• 30 Fail
• 40 Fail
• 50 Fail
• This means that
• Sensor sends analog values to Tmote
• Tmote communication is working
• Mainframes error-handling (exception, queue)
  working
• Sensor calibration is the only big thing left

7
Looking Forward

• Things weve learned
• Low-priced sensors are fishy
• Some datasheets are not documented well
• Other software in the system can be buggy
• We plan to
• Calibrate the sensor to minimize false negatives
• Improve error handling
• Finish strong!

8
Questions or Suggestions?

• Come with me if you want to live