Coveragebased Group Management Mechanism in Sensor Network

1
Coverage-based Group Management Mechanism in
Sensor Network

• Guoliang Xing
• Xiaorui Wang
• Yuanfang Zhang

2
Goals

• Basic goals
• Implement the function-based group formation
• Each group maintains sufficient coverage in a
  dynamic network with fault
• Advanced topics
• Sleep scheduling in a group
• Group maintenance

3
application
Sensors for cars Sensors for bicycles
4
application
Sensors for cars Sensors for bicycles
5
Assumptions

• Coverage
• Our assumption Any position in sensor network is
  sensed by at least n sensors (ex.
  acoustic/magnetometer sensors)
• Possible alternative User defined geographic
  region is sensed by at least n sensor (ex. N
  temperature/photo sensors per mile2)
• Sensing range
• Acoustic/magnetometer sensors circle

6
Sensing Range

• Acoustic/magnetometer sensor Omni-directional
• Sensitivity of Mica acoustic sensor
  Omni-directional, -45 4dB
• Assumption The range is circular
• R determined by sensitivity of sensor and sound
  transmission model

7
When n1, that means we need to make sure every
location in sensor network is within the sensing
range of at least one sensor.Now what we need
to do is using circles to cover the whole area
as sparse as possible. Some possible methods are
better than
Average overlap area comparison for each circle
(r1)
Obviously, the distance between nodes is also
larger for triangle.
8
Some Mathematic Considerations
Triangle covering Theorem If the other two
extended points are outside the starting points
sensing range and inside this triangle, as shown
at A and B area respectively. we can guarantee
these three sensing circle will at least cover
this triangle.
A
Points finding Guide in theory We will go from
starting point along the two edge without
crossing them and find two points in A and B. We
test length each hop and go back a step when
length is larger than . By this way we can
guarantee the coverage for the triangle
B
60
9
Example
East
10
Implementation considerations some revisions

• Implementation simplification
• Neighbor points merging to decrease density.
• Allow crossing edge to select points and compare
  which one is nearest to the ideal location
• Assumption We allow the possible uncovered area
  by this simplification, since we already consider
  the possible redundancy before.

11
Revised Coverage algorithm

• Algorithm
• Step1 Take a starting point randomly. This
  starting node will compute the six ideal
  locations and extend by six directions
• Step2 For each edge, the starting node will
  choose one node nearest to the ideal location as
  its group member.
• Step3 At each hop, the hop node will compute the
  length (length along the direction) to source
  node. If the length is larger than ,
  searching will stop and this node will compare
  itself and the previous node to get a nearest
  node to the ideal location

60
12
Example for revised algorithm
Here is the uncovered area due to revised
algorithm
East
13
Ripple Diffusing Policy
14
Ripple Diffusing PolicyCont.d

• Six Main ax.es
• Each node on main ax. looks for 2 children
• Main ax. direction
• 60 degree right-hand away from main ax.
• Each of other nodes looks for 1 child
• Parent-to-itself direction
• Covered area diffuses like a ripple until reaches
  the boundary of interested area

15
Extending to n2,3,4.

• By doing this algorithm once, we guarantee any
  location will be covered by at least one sensor.
• When n2, 3,N, we do this algorithm n times to
  make any location covered by n sensors.

16
Experiments

• Simulation experiments
• Coverage
• Empirical experiments
• Coverage
• Group formation time
• Overhead of group formation