CS perspective discussion

1
CS perspective - discussion
2
Sensing and range of applications

• Applications requiring
• Sensor based data collection
• Sensor network capabilities
• Wireless sensor network capabilities
• What are the data processing/management
  differences between them?
• Role of space/time scales

3
Role of models in SN

• Physical models
• Data driven models
• Failure models
• Data formats/Integrity of raw data
• Standardized data formats
• (Semi)-automated data quality checks
• Repurposing data sets

4
WSN characteristics

• Communication limited QoS (high latency, limited
  bandwidth, dropped packets, )
• Power limited energy supply -gt conservation
  crucial for long deployments
• Computation limited computing power and memory
  sizes
• Uncertainty in sensor readings signals detected
  at physical sensors have inherent uncertainty,
  sensor malfunction generates inaccurate data, bad
  design can lead to bad measurements

5
Mission of the WSN

• Just another data collection mechanism, or
• environment aware system?
• Sensors (or clusters of sensors) as single
  processing units
• Distributed data management system? (capable of
  answering declarative queries and providing data
  summaries, checking quality of collected data,
  fitting collection of models, )
• Cross-layer optimization issues how do the data
  management and communications layers interact?

6
Challenges

• Scalability
• Data multi-modality
• Sampling rates
• Which of these issues are still a challenge in a
  rich resource environment? How does the data
  collection and processing paradigm change?

7
System Needs

• What kind of tools, testbeds, methodologies are
  needed to ease the transition from network design
  to in-lab testing, to (long term) field deployment