Design and Implementation of a Middleware for Sentient Spaces

1
Design and Implementation of a Middleware for
Sentient Spaces

• Bijit Hore, Hojjat Jafarpour, Ramesh Jain,
  Shengyue Ji, Daniel Massaguer
• Sharad Mehrotra, Nalini Venkatasubramanian,Utz
  Westermann
• School of Information and Computer Science
• University of California, Irvine

2
Outline

• Rescue project Responsphere
• Multimodal sensor data processing
• SATware architecture
• Stream model
• SATRuntime Stream processing runtime
• SATLite Stream query and transformation language
• SATDeployer Operator deployment
• Privacy
• SATware Demo
• Related work

3
Rescue project and Responsphere

• Exploiting IT for effective response to
  unexpected events
• Responsphere a unique infrastructure for
  situation monitoring and awareness in the context
  of emergency response
• Includes more than 200 sensors of 10 different
  types deployed over the geographical space that
  covers about half of the UCI campus.

4
Multimodal sensor data processing Application

• SATware applications
• Surveillance and monitoring
• Emergency response
• Situation awareness
• Drills

5
Surveillance Application

• A building is equipped with multimodal sensors
• Video cameras. RFID readers, Motes,..
• Specific event should be detected and
  notifications should be generated

6
Multimodal sensor data processing Challenges

• Abstraction
• Entities and spaces
• Virtual sensor concept
• Multimodal stream processing
• Heterogeneous stream sources e.g., video cam,
  RFID, mote, ..
• Scalability
• Extensibility
• Mobility
• Privacy

7
SATware

• A distributed middleware for capturing
  semantically meaningful events
• Provides higher level of semantic abstraction
  compared to sensor middleware
• Two level stream processing
• Semantic event stream processing
• Sensor data stream processing

8
SATware stream processing model
9
SATware architecture Stream model

• Data model
• A stream is a flow of records ltt , C ,
  Confidencegt
• t time stamp
• C content
• Confidence probability of the correctness of
  tuple content
• lt 121025 , PersonID 7 , 0.9 gt
• lt 113231 , RoomOccupancy 4 , 0.75 gt

10
SATware architecture Stream model

• Processing model
• Operators
• Input, output buffer
• Operator types
• Image processing operators
• Join operators
• Synchronization operators
• Filtering operators

Oi
11
SATware architecture Stream model
Sample image processing operator
lt ti , Door Status gt
lt ti , Jpeg image gt
Oi
Sample synchronizer operator
lt ti , PersonID gt
lt ti , PersonID , Door Statusgt
Oi
lt ti, Door Statusgt
Sample join operator
lt ti , PersonID gt
lt ti , PersonEntery gt
Oi
lt ti, Door Statusgt
12
SATware architecture
13
SATware architecture SATRuntime

• Main components
• Set of processing nodes
• SATRuntime component
• One SATRuntime on each processing node
• Interface for operators (running, accepting,
  migrating, connecting,)
• Operators as mobile agents
• Input stream, output stream buffers
• Operator connection interface
• Operator migration

14
SATware architecture SATRuntime

• SATRuntime infrastructure directory
• Available streams
• Sensor and Operator repository
• Sensor management
• public boolean addSensorInstance(SensorInstance
  sensorInstance)
• Sensor.setLocation( Location l )
• Operator management
• public boolean addOperatorInstance(OperatorInstanc
  e operatorInstance)
• Deployment status
• Operator graph
• Operator location

15
SATware architecture SATLite

• SATLite Defines the query plan
• Set of interconnected operators fed by sensors
• SATLite provides a query plan by translating the
  query into a graph
• Select t , PersonID from DoorStatus and PersonID
  where Person.t DoorStatus.t and
  DoorStatusOPEN

lt ti , PersonID , Door Statusgt
lt ti , PersonID gt
Oj
Oi
lt ti , PersonID gt
lt ti, Door Statusgt
16
SATware architecture SATDeployer

• Translating a query plan into a deployment plan
• Mapping query plan nodes into processing nodes
• Minimizing query processing cost
• Network cost
• Computing cost
• Query evaluation time
• Optimizations
• Operator reuse
• Load balancing

17
SATware architecture Privacy

• Hiding individuals identity
• Revealing individuals identity when a predefined
  policy is violated
• Sensors are trusted components
• Adversary is an inquisitive but non-interfering
  (passive) entity that can observe all stored data
  and intercept all communication between two
  components of the system.

18
Sentient building demo

• A pervasive space equipped with sensors
• Video cams, RFID readers, Motes, Mobile sensing
  platforms,
• SATRecorder a building surveillance application
• Recording video feed based on timed schedule or
  event detection
• Coffee room scenario
• Sensors 2 video cams, RFID reader
• Sample policy if pot is empty and burner is on
  record video

19
Sentient building demo
20
Sentient building demo
21
Sentient building demo coffee room

• Event definition
• Select PersonID from KitchenEvent where
  CoffeeLevel EMPTY and PotInPlace TRUE and
  BurnerStatue ON
• Network model and query plan

22
Sentient building demo coffee room
23
Related work

• Data stream processing
• Aurora, STREAM, TelegraphCQ
• Sensor networks
• Multimedia stream processing
• IBM S3, MedsMan, IrisNet, Smart Camera Network

24
Conclusion

• We proposed
• Challenges in sensor stream processing
• SATware A middleware for sentient spaces
• A surveillance application based on SATware
• Future plan
• Expanding sensor deployment
• Sophisticated event detection
• Improving SATLite and SATDeployer
• Integrating more sophisticated privacy preserving
  techniques