Risks & Mitigation Strategies

  1. Denial-of-Service Attacks
    • What happens if an attacker pretends a thousand Bluetooth devices are suddenly active?
    • Or if the host throws a party and a thousand people come over?
    • Whose settings do we load?
    • Solution: Limit the interface to a single-user
  2. Manual Overrides
    • What if someone who doesn’t know about Maestro changes some settings?
    • Solution: Design Maestro so that it understands that device changes can happen via an external agent
  3. Bandwidth Limitations
    • Transmission of audio/video data over Bluetooth is painful.
    • Solution: Use lossy compression, because the important part is that we get an idea of what’s going on. It doesn’t matter if the signal at 320 Hz is completely gone as long as we hear almost everything else.
  4. Unauthorized Use
    • Say two people set up Maestro at home. What’s to keep one person from going over the other person’s house and controlling everything?
    • Solution: Use both Bluetooth and RFID for two-factor authentication

Error Handling

Test Cases

    Test Range and Latency of IR Sensors:

     

    Independent Variables: Supply Voltage of Sensors, Distance of Sensors, Lighting Conditions.

    Dependent Variables: Latency and presence of the returned signal after the pulse is sent out.

     

    Plot latency as a function of distance and the farthest point of being able to sense the beam for several different supply voltages (5v, 7.5v, 9v, 12v, 15v, 18v)

     

    Determine the optimal supply voltage for a standard door width, and determine the latency of the pulse for that voltage and distance.

     

    (The IR sensors fire once every 20 ms, so for an object to escape the notice of the IR sensors it must pass through the plane of the doorway in that time, so it depends on how long the object is.  For a typical human being (let¡¯s say a minimum depth of 6 inches), he would have to be moving at 17 miles per hour.)

     

    To show that user¡¯s distance from gumstix can be measured by Bluetooth signal strength

    Test setup: gumstix + bluetooth, phone, ruler, sources of interference (devices competing in 2.4 GHz spectrum, walls to weaken signal strength, other Bluetooth devices connected to gumstix)

    Metrics: distance and orientation of phone relative to BT antenna, measured Bluetooth signal strength

    Workload: gumstix connects to phone, takes measurements for 20 seconds, computes average signal strength

    Parameters: class of Bluetooth devices, output power, amount of interference

    Test Run: place phone at certain distance with Bluetooth on, check level of interference, start signal-measuring program on gumstix, record estimate of signal strength until no changes in signal are detectable (maximum distance)

     

    Measure latency of turning on lamp from when X10 controller is activated

    Test setup: gumstix + X10 controller, incandescent lamp, clear channel (no noise) from controller to lamp over power line, room is dim enough for us to see the lamp

    Metrics: response time for the light to turn on (end time is when light is on at full brightness, start time right before the first data is sent to the X10 controller from the gumstix)

    Workload: lamp¡¯s job to deliver brightness to the room

    Parameters: (final) brightness of the lamp

    Test run: lamp is plugged into X10 receiver, X10 controller (connected to gumstix) is plugged into the same power circuit, turn X10 controller on, time how long it takes for user to notice the light is on

     

    Measure range and field of view of motion detector, latency of detection

    Test setup: motion detector + gumstix, something to measure the angle + distance (protractor + ruler?), tape, active human that the motion detector can detect, large area to experiment in, static environment so motion detector only detects test subject

    Metrics: what areas the test subject can trigger the motion detector in (measure range and angle from motion detector¡¯s primary axis), whether the motion detector is triggered, how long it takes from person¡¯s first movement to trigger of motion detector signal

    Workload: person tests at distances at multiples of 1 foot from motion detector up to 50 feetb, intervals of 15 degrees, up to 180 degrees

    Parameters: stature, type of person¡¯s action

    Test run: person jumps up and down without changing position relative to motion detector, motion detector detects an action and outputs something measurable by the gumstix

     

Experimental Evaluation