Optimal Data Compression and Forwarding in Wireless Sensor Networks

1
Optimal Data Compression and Forwarding in
Wireless Sensor Networks

• Bulent Tavli, Mehmet Kayaalp, Ibrahim E. Bagci
• TOBB University of Economics and Technology
• Ankara, Turkey

2
Goals

• Maintain balanced energy consumption among
  sensors
• Increase network lifetime
• Focus on whole network rather than individual
  nodes
• Exploit data compression
• Explore different strategies for mitigating
  sensor network hotspots

3
Transmission Scheduling
In a many-to-one (converge-cast) multi-hop
wireless sensor network, how should we schedule
transmissions so as to balance energy usage and
maximize lifetime?

4
Direct Transmission

High energy drain in the furthest nodes
5
Next Hop

High energy drain in the closest nodes
6
Split Transmissions

Will a scheme like this help?
7
Problem Definition

• Given
• Sensor locations
• Power model
• Traffic generation pattern
• Initial energy distribution
• Goal
• Determine optimal flow pattern to maximize
  network lifetime
• Solution
• Linear programming

8
Models

• Power model

• Compression model

9
Linear Program for flow balancing
10
Linear Program for flow balancing and data
compression
11
Strategies

• NCFB (No Compression and Flow Balancing)
• Only flow balancing
• MCFB (Mandatory Compression and Flow Balancing)
• All nodes compress all of their data
• Flow balancing
• OCFB (Optimal Compression and Flow Balancing)
• Nodes compress their data and balance the flow on
  the network jointly

12
(No Transcript)
13
Example 1 All nodes compress
Pcp ?, node-separation 15m
14
Example 2 No compression at all
Pcp 10?, node-separation 15m
15
Example 3 Some compression
Pcp 10?, node-separation 25m
16
Example 4 All nodes compress
Pcp 10?, node-separation 80m
17
Conclusions

• Data compression is becoming an integral part of
  in-network data processing
• Allocate energy budget on compression and
  forwarding optimally
• Linear Programming
• Avoid data compression
• Small network high compression energy
• Partial data compression
• Large network high compression energy
• Small network low compression energy
• Compress most of the data
• Large network low compression energy
• For all parameter space jointly optimizing data
  compression and load balancing results in maximal
  network lifetime

18

• QA