EE5369: Efficient Data Aggregation

1
EE5369 Efficient Data Aggregation

• To study implement data aggregation
  algorithm(s) in a static wireless sensor network
  using MICA motes using NesC on the TinyOS
  platform.

Objective
Team Muthiah Annamalai, Darshan Shrestha
03/30/06, UTA
s
2
Introduction
EE5369 Efficient Data Aggregation
Principles
Techniques

• Energy efficient source coding
• Exploit spatial temporal correlation
• CostTx gt CostCalc
• Lossless, Lossy

• Distributed Source Coding
• Wavelet processing
• A/µ law companding
• RLE, Huffman coding

03/30/06, UTA
3
Background Information
EE5369 Efficient Data Aggregation
Principles
Implementation Facts

• Best data compression
• Inf' Theory, RH(x)
• Physical Process as Correlated source.
• Prefix free codes.
• Unique decoding.
• Fast encoding LUT O(1).
• Decoding polynomial time
• Variable Bitrate scheme.
• Application dependent factors
• Latency, Bitrate, Quantization

• Slepian-Wolf theorem on 'Noiseless coding of
  correlated sources' proves existence of rate
  regions.
• R1 H(xy)
• R2 H(yx)
• R1R2 H(x,y)
• Pradhan, Ramachandran DISCUS Distributed Source
  Coding, scheme.
• Research on using channel codes for correlated,
  distributed source coding.
• Network Coding
• Wavelet processing, Mutiresolution schemes.

03/30/06, UTA
4
Expected Results
EE5369 Efficient Data Aggregation
Performance Analysis
Scenarios

• Latency Codecs
• Bit rate
• Quantization error
• Energy utilization

• Temperature monitoring
• Intrusion detection.
• Application based source code
• CodeF(Latency,BitRate,
  Error,Energy)

03/30/06, UTA
5
Merits
EE5369 Efficient Data Aggregation

• Focus on new recent past 4 years research.
• DSN and DSC are energy savers.
• Many simulations. Lack of experimental data.
• Will study suitability of source coding for
  applications.
• Excellent opportunity to utilize communication
  background, and system integration.

03/30/06, UTA
6
Tasks
EE5369 Efficient Data Aggregation

• Darshan
• Mica, Sensor data calibration, control.
• Encoding of few schemes.
• TinyOS communication protocols.
• Sensor perturbation.
• System Integration

• Muthiah
• Source codec Simulation testing
• TinyOS source codecs.
• Data log' timestamp.
• Mote timing and sync.
• System Integration

03/30/06, UTA
7
References
EE5369 Efficient Data Aggregation

• Thomas Cover, Joy Thomas, "Elements of
  Information Theory"
• D.Slepian, JK Wolf, Noiseless coding of
  correlated information sources, IEEE
  transactions on Information Theory, 1973
• Jugoslava Acimovic,"Efficient Distributed
  Multi-resolution Processing for Data gathering in
  sensor networks",ICASSP 2005
• Jugoslava Acimovic,"Adaptive Algorithms for
  power-efficient data gathering in sensor
  networks", IEEE 2005
• Ruixin Niu,"Fusion of Decisions Transmitter over
  Rayleigh fading channels in wireless sensor
  networks", IEEE 2006

03/30/06, UTA
8
QAThank You.
EE5369 Efficient Data Aggregation
03/30/06, UTA