Networked Embedded Systems

1
Networked Embedded Systems

• Kamin WhitehouseCS 696
• 09/05/07

2
Technology Trends

• Sensors/actuators entering human habitats
• 32 million homes have security sensors
• 5 million homes have X10 devices
• Estimated 20 million ZigBee devices by EOY
• We will need
• To program them
• To debug them
• Privacy preservation
• Data stream processing

3
Macro-programming
RPC (marionette)
for i 1100 val nodei.light.read() date
nodei.time() if val lt 50 printf(Error!)
else hashtable.put(i,val,date) end pda.di
splay(val) end
node
node
pda/server
node/server
pda
4
Macroprogramming
Automatic Decomposition

• for i 1100
• val nodei.light.read()
• date nodei.time()
• if val lt 50
• printf(Error!)
• else
• hashtable.put(i,val,date)
• end
• pda.display(val)
• end

node
node
pda/server
node/server
pda
5
Macroprogramming
Automatic Decomposition
for i 1100 val nodei.light.read() date
nodei.time() if val lt 50 printf(Error!)
else hashtable.put(i,val,date) end pda.di
splay(val) end
node
node
server
server
pda
6
Macroprogramming
QoS Satisfaction
L lt 350

• for i 1100
• val nodei.light.read()
• date nodei.time()
• if val lt 50
• printf(Error!)
• else
• hashtable.put(i,val,date)
• end
• pda.display(val)
• end

node
node
server
server
pda
7
Macroprogramming
QoS Satisfaction
L lt 350

• for i 1100
• val nodei.light.read()
• date nodei.time()
• if val gt 50
• printf(Error!)
• else
• hashtable.put(i,val,date)
• end
• pda.display(val)
• end

node
node
server
server
pda
8
Macroprogramming
QoS Satisfaction
L lt 350

• for i 1100
• val nodei.light.read()
• date nodei.time()
• if val gt 50
• hashtable.put(i,val,date) else
• printf(Error!)
• end
• pda.display(val)
• end

node
node
server
server
pda
9
Macroprogramming
QoS Satisfaction
L lt 350

• for i 1100
• val nodei.light.read()
• date nodei.time()
• if val gt 50
• hashtable.put(i,val,date) else
• printf(Error!)
• end
• pda.display(val)
• end

node
node
node
pda
pda
10
Macroprogramming
QoS Satisfaction
L lt 350

• for i 1100
• val nodei.light.read()
• date nodei.time()
• if val gt 50
• hashtable.put(i,val,date) else
• printf(Error!)
• end
• pda.display(val)
• end

node
node
node
pda
pda
11
Macroprogramming
QoS Satisfaction
L lt 350
for i 1100 val nodei.light.read() date
nodei.time() if val lt 50 printf(Error!)
else hashtable.put(i,val,date) end pda.di
splay(val) end
node
node
server
server
pda
12
Outline check

• Programming
• Debugging
• Privacy
• Data Processing

13
Debugging
14
Debugging
Program
MCU
Add a trap
Debugger
15
Debugging
16
Debugging

• Break
• Step
• Watch
• Backtrace
• Etc
• 30KB of program memory
• 1KB of RAM

17
Outline check

• Programming
• Debugging
• Privacy
• Data Processing

18
Privacy Preservation

• Home or away
• Awake or asleep
• Bathroom usage
• Kitchen usage
• Showering, toileting, washing
• Cooking hot food or preparing cold food

19
FATS Attack
Bathroom
Kitchen
Living Rm
20
FATS success across 4 homes
21
Preserving Privacy
Complete Privacy
Bayes Rule
Requirement For Privacy
22
Counter Attacks

• Periodic Transmissions
• Assumes tolerable latency bound L
• Does not work with real-time or high bandwidth
  requirements
• Consumes bandwidth
• Consumes power
• Random Delays
• Exploit L with lower power bandwidth
  requirements
• Still assumes L

23
Counter Attacks

• Mask fingerprints in hardware by varying features
  for each transmission
• Arms race scenario, unable to predict features
  used by an adversary
• Not supportable by current hardware
• Does not affect inference of sleep and home
  occupancy variables

24
Counter Attacks

• Increasing Packet loss ratio by
• Reducing transmission power
• Introducing RF attenuators

25
Outline check

• Programming
• Debugging
• Privacy
• Data Processing

26
Data Stream Processing

• Can the user infer desired info without making
  strong assumptions?

27
Data labeling

• Can we detect patterns without assumptions?
• Major time saver

28
Data Labeling
29
Results
30
Results
31
Data Sharing

• Personal sensors are prevalent
• Homes
• Cars
• Phones
• Shoes, etc
• Goal create infrastructure for sharing data
  creating value

32

• Shopkeepers publish data
• people in front of store
• people coming into store
• Credit card purchase info, etc
• Entrepreneurs provide service
• Effect of weather, concerts, etc on business
• Effect of advertising
• How to increase conversion rates, etc
• Plus
• Overall activity downtown
• Value of commercial real estate
• Effect of vehicular traffic on businesses, etc.

33
Search

• Search is key to data sharing
• PageRank
• StreamRank mines the WWSW and creates links
  between data streams
• Correlation
• Ownership
• Browsing, etc

34
Thanks

• Kamin Whitehouse
• Computer Science Department
• whitehouse_at_cs.virginia.edu