Application Requirements Breakout I: Vision

1
Application Requirements (Breakout I Vision)

• Coordinator Alex Labrinidis
• Scribe Cyrus Shahabi

2
Application examples

• Military application example
• Every soldier is a sensor
• Objective situation awareness
• Where they are in the environments and
  detecting/reporting the threats
• Need Geo-sensor network (local filtering of
  information), autonomous vs. centralized/hierarchi
  cal decision making

3
Application examples

• Intelligent transport application example
• Motivation time wasted in traffic safety of the
  drivers EU spends
• Each vehicle is a (set of) sensor ? no battery
  issue here
• Broadcast, vehicle-to-vehicle (p2p)
• Queries ABS triggered in one car propagates to
  other cars (events) collecting usage data to
  suggest/predict destinations and routes (closest
  gas station) traffic congestion
  detection/control (predict travel time) fleet
  management
• Moving maritime vehicles
• Re-routing ships
• Container tracking

4
Application examples

• Hybrid of military and transportation example
  unmanned autonomous vehicle (DARPA challenge)
  UAVs
• Objective sensing environment, impacting the
  battlefield

5
Application examples

• Phenomena detection tracking example
• Phenomena itself is moving (brushfire)
• Geo-sensor networks
• RFID-based sensors and WallMart type applications
• Real-time location-based services
• Crawling and searching live sensor data
• Available parking spaces
• How delayed is the bus
• Integration of deep-web with real-time sensor web

6
Evaluation of applications

• User interface issues
• Privacy issues location of people with mobile
  device
• Security and trust issues
• Who are the users
• How much autonomy? (safety)
• Metrics

7
Generalization Common Needs of Applications

• A database that handle
• Past history data
• Now current reading of sensor data
• Future predictions
• Distributed data
• Location time data are important (how to store,
  index, query efficiently)
• Uncertainty in data
• Self-configurable, support mash-ups

8
Where to start?

• User-inspired research (application-aware) vs.
  pure fundamental research
• A case for the former the applications are too
  diverse that we need to define the application to
  focus research
• A case for the latter research leaps
• Instead of starting with a visionary new
  application, try an incremental approach by
  changing the current popular applications.
• e.g., restaurant example (how many people
  waiting)
• Enable a technology and let other people come up
  with the successful apps (YouTube) risk can
  enable bad things!
• In-between well defined user base, create an
  application platform and give out APIs,
  google-maps

9
What is Unique in Intelligent Transportation
Systems?

• People stuck in traffic, Congestions
• Predictable travel time, travel time is quality
  time
• 4.6 billion Euros of funding in ITS and
  navigation systems
• ITS system Need to be online
• Dont have battery problems

10
Application Requirements (Breakout II)

• Alex, Christian, Gerry, Le, Ling, Mohamed,
  Vassilis, Vincenzo, Walid

11
Application Requirements Refining the Vision

• What is mobile?
• Sensors
• Data that we are measuring (or Phenomena)
• User can be mobile or not
• Can you control the sensor mobility or not?
• Processing mobility
• What additional challenges/benefits does mobility
  add to sensors?
• Challenge Sensor connectivity
• Benefit Can supplement missing readings
• Benefit Can focus more sensors in the
  interesting areas
• Benefit Can do better tracking with mobile
  sensors
• Benefit Much more expensive to deploy stationary
  sensors
• E.g., Sensors to count number of cars

12
Requirements Questions

• What is being sensed?
• What is mobile?
• What do we need to capture (w.r.t. the Time
  Dimension)
• Forensic (past), Real-time (present), and
  Predictive (future)
• Where will the data be stored?
• Both In-network (distributed) vs. out-of-network
• Processing
• Challenges

13
App.1 Soldier as a Sensor

• What is being sensed?
• Threats, soldiers health condition, location,
  their perception of the threat space
• What is mobile?
• Sensor Is mobile
• Some tracked objects/phenomena are mobile
• Consumers/users may be mobile
• What do we need to capture (w.r.t. the Time
  Dimension)
• Forensic (past), Real-time (present), and
  Predictive (future)
• Where will the data be stored?
• Both In-network (distributed) and out-of-network
• Processing
• Joining sensor data with other sensor/non-sensor
  data
• Locality awareness
• Challenge Sensor (and computing) mobility
  control based on application semantics
• Challenge Uncertainty, missing data, power,
  connectivity
• Challenge Answering queries when data is stored
  in in-network in a mobile unit (variability in
  the position of the sensor)

14
App. 2 Intelligent Transportation Systems

• What is being sensed?
• In-vehicle sensors Vehicle location, state
  (contents), number of passengers (people in a
  bus), belts
• In-road sensors Number of vehicles (in-road
  sensors), Road condition, sense the environment
• What is mobile?
• In-vehicle sensors
• Phenomena are mobile (e.g., congestion)
• Consumers/users may be mobile
• What do we need to capture (w.r.t. the Time
  Dimension)
• Forensic (past window last n miles or last few
  seconds), Real-time (present), and Predictive
  (future)
• Data mining of past data
• Where will the data be stored?
• Both In-network (distributed) and out-of-network
• Processing
• Real-time operations, e.g., filtering, and
  monitoring events
• Sense and Respond (How one reacts based on sensed
  data?)
• Challenge Computing mobility control based on
  application semantics
• Challenge Add network infra-structure in
  real-time whenever and wherever it is needed
• Challenge Understand the actual problems -
  interdisciplinary
• Challenge Large scale projects and
  prototypes/testbeds that help with the
  understanding

15
Application Requirements (Breakout III Specific
Research)

• Alex, Christian, George, Gerry, Mohamed, Ouri,
  Panickos, Reed, Vassilis, Walid, Wang-Chien

16
Challenging Applications

• Meta advise Find applications where there is
  potential for bringing high value, with
  challenging requirements, and with a business
  partner
• Medical applications
• Sports
• The elderly
• Chronic illnesses (e.g., diabetes, allergies)
• Wellness
• Disaster and emergency management
• Soldier as a sensor
• Intelligent transport systems

17
Research Challenges

• Expanding Internet-like search to sensor data
• Integration of the web, the deep web, and the
  sensor web
• Highly volatile data sampling of continuous
  variables stream data
• Mobility and varying QoS poses challenges
• Context-awareness
• situation awareness
• Push functionality will be prevalent
• The system establishes your context automatically
  and uses this to improve functionality
• Enabling unobtrusiveness, disappearing computer
• Enables better system performance
• Important in many applications
• Soldier is a sensor, ITS, medical
• Profiles as relatively static context
• Preferences

18
Research Challenges

• Purpose-driven data reduction
• Purpose specification language
• Time spectrum of uses ranging from real-time to
  off-line
• Optimization opportunities
• Ad-hoc and continuous query support
• Mobile multi-query optimization
• Dynamic, adaptive aspects
• In-network data reduction
• Plug and Play/self-everything (configurable,
  etc.)
• Ease of use, complexity of systems
• Adaptive, learning behavior
• In applications, user interfaces

19
Cross-Cutting Challenges

• Privacy, security, trust
• Inaccuracy of data sampling of continuous
  variables
• Inaccessibility of data
• Missing data
• Tracking of mobile phenomena/objects (fire,
  vehicle)
• Object correspondence/identification
• Understand the real problems!
• Geo-location
• Post-analysis, forensic analysis, historical data
  analysis

20
Research Challenges Summary

• Top 3 challenges
• Purpose-driven data reduction
• Context-awareness
• Searching the sensor web
• Really cool stuff
• Combining Internet search with sensor data
• Improving personal health and quality of life