Sensor Networks Deployment Using Flip-based Sensors

1
Sensor Networks Deployment Using Flip-based
Sensors

• Sriram Chellappan, Xiaole Bai, Bin Ma and Dong
  Xuan
• IEEE International Conference on Mobile Ad hoc
  and Sensor Systems Conference, 2005. Nov. 7,
  2005 Page(s)291 - 298
• Presented by Jeffrey

2
Outline

• Introduction
• Related Work
• Mobility Model and Problem Definition
• Proposed Solution
• Performance Analysis
• Conclusions and Future Work

3
Outline

• Introduction
• Related Work
• Mobility Model and Problem Definition
• Proposed Solution
• Performance Analysis
• Conclusions and Future Work

4
Introduction

• To improve the sensing coverage if there exist
  holes
• In the literature
• If a sensor chooses to move to a desired
  location, it can do so without any restriction
• NOT realistic

5
In Practice

• A class of Intelligent Mobile Land Mine Units
  (IMLM) to be deployed across battlefields have
  been developed by DARPA
• mobility in the units is restricted to only a
  hopping mechanism
• Each IMLM unit carries onboard fuel tanks and a
  spark initiation system to propel the hop

6
Outline

• Introduction
• Related Work
• Mobility Model and Problem Definition
• Proposed Solution
• Performance Analysis
• Conclusions and Future Work

7
Related Work

• Dynamic Coverage Maintenance Algorithms for
  Sensor Networks with Limited Mobility
• PerCom 2005

8
Minimum Distance Lazy (MDL)
9
Cascaded DCM Scheme
10
Outline

• Introduction
• Related Work
• Mobility Model and Problem Definition
• Proposed Solution
• Performance Analysis
• Conclusions and Future Work

11
Mobility Model and Problem Definition
12
Optimal Movement Plan
13
Outline

• Introduction
• Related Work
• Mobility Model and Problem Definition
• Proposed Solution
• Performance Analysis
• Conclusions and Future Work

14
Proposed Solution

• A centralized node (a Base station) collects
  information about the number of sensors in the
  regions
• Propose a minimum-cost maximum-flow based
  solution that is executed by the Base-station
  using the region information
• The output of our solution is a movement plan
  (which sensors should move and where) for the
  sensors

15
Virtual Graphs
16
Virtual Graphs -Construction when Fd and Rd
17
Virtual Graphs -Construction when Fd and Rgtd
18
Proposed Algorithm

• We first determine the value of the maximum flow
  in from all Source vertices to Hole vertices
• Using the Edmonds-Karp algorithm

19
To Get The Minimum Cost Flow

• Use Solving minimum-cost flow problems by
  successive approximation
• It works by starting to find an approximate
  solution and then iteratively improving the
  current solution

20
Outline

• Introduction
• Related Work
• Mobility Model and Problem Definition
• Proposed Solution
• Performance Analysis
• Conclusions and Future Work

21
Performance Analysis

• Metrics
• Coverage Improvement (CI)
• CIQo-Qi
• J
• Optimal number of flips as determined by our
  solution
• Flip Demand (FD)
• FDJ/CI

22
Evaluation Environment

• Field Size
• 150 ? 150
• Region Size
• 10
• ?
• change from 0 (uniform distribution)
• to 4 (highly concentrated at the center of the
  field)

23
Sensitivity of CI to F when R varies
24
Sensitivity of CI to F when ? varies
25
Sensitivity of FD to F
26
Outline

• Introduction
• Related Work
• Mobility Model and Problem Definition
• Proposed Solution
• Performance Analysis
• Conclusions and Future Work

27
Conclusions and Future Work

• Proposed a minimum-cost maximum-flow based
  solution to optimize sensor network deployment
  using flip-based sensors
• Studied the sensitivity of performance to flip
  distance, under different initial deployment
  scenarios

28
Future Work

• Considered flips in increments of a basic unit
  (d)
• Discrete case
• Try to relax this in order to handle continuous
  mobility, although the overall movement distance
  (d) is still limited

29
Comments

• Strong
• Elegantly convert a real world scenario to a math
  model and solve it
• Weak
• Is the sensing coverage model realistic?