HighResilient, EnergyEfficient Multipath Routing in Wireless Sensor Networks

1
High-Resilient, Energy-Efficient Multipath
Routing in Wireless Sensor Networks

• Deepak Ganesan UCLA
• Ramesh Govindan UCB
• Scott Shenker ACIRI
• Deborah Estrin UCLA
• Mobile Computing and Communications Review 2001

2
Outline

• Introduction of Sensor Networks
• Multipath Routing Algorithm
• Measurement Factor
• Performance Result
• Conclusion

3
Introduction of Sensor Networks

• May deploy in inaccessible environment.
• Limited Power, Computational Capacities, Memory.
• Node are prone to failures. (environment / power)
• Topology changes very frequently.
• May not have global ID. ( Ex IP address)

4
Introduction of Sensor Networks (cont.)

• Protocols and algorithms must possess
  Self-organizing capability
• Cooperative effort. ( Sensor node fitted with
  on-board CPU )
• Application areas
• Health , military , security
• Ex wearable wireless network

5
Sensor Network
6
Design Factor

• Fault Tolerance
• Scalability
• Production costs
• Operation Environment
• Network Topology
• Hardware constraint
• Transmission media (Channel)
• Power Consumption

7
Multipath Routing Algorithms

• Directed Diffusion (Diffusion)
• Disjoint multipaths routing (D-MPR)
• Braided multipath routing (B-MPR)
• Combining multipath routing and data-centric
  routing with localized path setup.

8
Directed Diffusion

• Data generated by sensor is named using
  attributed value pairs.
• A sensing task is disseminated throughout the
  sensor network as an interest for named data.
• This disseminated sets up gradients within the
  network designed to draw events.

9
Directed Diffusion (cont.)

• Events start flowing toward the originators of
  interests along multipaths.
• The sensor network reinforces one, or a small
  number of these paths.

10
Directed Diffusion (cont.)
Periodically broadcast
11
Directed Diffusion (cont.)
Primary path
12
Directed Diffusion (cont.)
Periodically send reinforcement
13
Disjoint multipaths routing

• Construct a small number of alternate paths that
  are node-disjoint with primary path.

14
D-MPR (cont.)

• K node disjoint
• Construct the primary path P between source and
  sink
• The first alternate disjoint path P1 is the best
  path node-disjoint with P
• The second alternate disjoint path P2 is the best
  node disjoint with P and P1, and so on.

15
D-MPR (cont.)
16
D-MPR (cont.)
After receive data
17
Braided multipath routing

• D-MRP can be longer and expend more energy.

18
Perfect B-MPR
19
Measurement Factor

• Resilience
• Complete failure of multipath (high is good)
• Maintenance overhead
• Power consumption
• Failure mode
• isolated node failures (single node)
• Patterned failures (area)

20
Simulation parameter

• Uniformly distributing nodes on finite plane of
  dimension 400 M square.
• Transmission radius 40 M
• Density ( Number of Node)
• Spatial separation ( hop count )
• Failure probability Pi

21
Simulation Result
High maintenance overhead
High resilient
Illustrating the energy vs resilience trafeoff
22
Simulation Result (cont.)
High maintenance overhead
23
Simulation Result (cont.)
The impact of density on maintenance overhead
24
Simulation Result (cont.)
The impact of density on maintenance overhead
25
Simulation Result (cont.)
The impact of failure probability on resilience
in isolated node failure
26
Simulation Result (cont.)
The impact of density on resilience in failure
node failure
27
Conclusion and Future Work

• The energy cost of alternate disjoint paths
  depends on the network density.
• D-MRP give us independence fail on primary path
  without impacting the alternate path.

28
Conclusion and Future Work

• Primary path can be recovered without invoking
  network wide flooding.
• Resilience and energy is a trade off.
• Braided multipaths expend only 33 of the energy
  of disjoint multipaths.
• Braided multipaths have a 50 higher resilence to
  isolated failures than disjoint multipaths.

29
Conclusion