RedBlue Wireless Sensor Network Simulation

1
Red-Blue Wireless Sensor Network Simulation
Project Report Presentation submitted to the
Faculty of the Department of Computer
Science, Old Dominion University, in Partial
Fulfillment of the requirements for the Degree
of Master of Science

• by Ekaterina Shurkova
• under supervision of Dr. Stephan Olariu

2
Agenda

• Introduction
• The goal of the project
• Red-Blue Wireless Sensor Network
• Implementation
• Simulation results
• Collisions Problem, Channels
• Channels Utilization
• Conservativity
• Sensors with Simple Memory
• Conclusions
• Future Work
• Acknowledgements
• References

3
Introduction

• massively deployed sensor network population of
  small entities e.g. colonies of cells
• emerging behavior of large groups of sensor
  nodes
• field 10 by 10 units, 20,000 sensor nodes
• communication between sensors short-range radio
• each sensor gathers the information from its
  neighborhood (listening and transmitting modes of
  the radio)
• each node decision is made based on this local
  information
• local stability leads to a global stability
• collective behavior vs. centralized control

4
The goal

• The goal of the project is
• to deploy red and blue sensor nodes
• and to let them collectively distribute
  themselves to get blue and red color sensors
  uniformly spread out through the field
• is it possible to observe an emerging behavior?
• Whats the point?
• relation wolfs-sheep - an environmental
  stability
• too many wolfs problem
• too many sheep problem

5
Illustrating Model

• Environment field of grass, lawn
• Sensor Nodes smart sprinklers which are able
• to sense local area
• to analyze humidity
• to make a decision
• turn myself on
• turn myself off

6
Illustrating Model (2)
The Goal
Stability problems
Too many sprinklers ON
Too many sprinklers OFF
7
Sensor Node

• Sensor
• Processor
• Transceiver device
• Genetic material
• color red or blue
• status phase of life cycle period
• timing of each life cycle phase
• local clock
• ability to calculate neighborhood coloring
  scheme
• ability to change color based on neighborhood
  coloring scheme
• transmission range radius of radio
  transmission
• conservativity.

8
Sensor Life Cycle
9
Sensor setup
color red, blue time_to_sleep 60, 61, 62
units time_to_L/R 3, 4, 5
units time_to_L/B 3, 4, 5
units time_to_T 3, 4, 5
units status sleep, L/R, L/B,
transmit localClock 0 TR 1
10
Sensor setup (2)
color blue time_to_sleep 60, 61,
62 time_to_L/R 3, 4, 5 time_to_L/B
3, 4, 5 time_to_T 3, 4, 5
gt 61 units
gt 4 units
gt 3 units
gt 5 units
status sleep localClock 19
randomly seats itself on a life cycle timing
interval gt
11
At the Deployment Moment
status sleep localClock 13
status L/R localClock 2
status L/B localClock 1
status sleep localClock 1
12
Sensors Interaction
of heard red neighbors is accumulated
of heard blue neighbors is accumulated
decision on changing color is made
sensor transmits its color
13
Sensors Interaction (2)
Communication (Transmission) range of the sensor
compared to the entire deployment field.
14
Neighborhood Coloring Scheme Balance

• once sensor accumulated info about red and blue
  neighbors, it calculates the balance in coloring
  scheme within its transmission range
• the balance is the number between -1 and 1, since
  
• if blues 0 (no blue neighbors), then balance
  -1
• if reds 0 (no red neighbors), then balance
  1
• if balance 0, then the neighborhood is ideally
  stable
• since sensor gets the information from awake
  transmitting sensors, the information is not
  accurate enough some of the neighbors are
  listening and some of them are sleeping

15
Conservativity

• to handle such an inaccuracy a sensor is
  conservative, if balance is in range -0.25
  0.25 sensor assumes that the neighborhood is
  stable enough and does not change the color
• if balance gt 0.25, it means that there are too
  many blue sensors within its transmission range
  and if the sensor is blue, it should change a
  color
• if balance lt -0.25, it means that there are too
  many red sensors within its transmission range
  and if the sensor is red, it should change a
  color
• otherwise the decision is not to change a color

16
Implementation

• Microsoft Visual Studio 6.0, Visual C
• class environment
• class sensor
• struct message

17
Graphical Interface
18
Simulation Results Random Deployment
Plot 1a. Randomly deployed 20,000 sensors.
19
Simulation Results Random Deployment (2)
Plot 1b. Randomly deployed 20,000 sensors
(different experiment).
20
Simulation Results Red superiority
Plot 2a. 15,000 red and 5,000 blue sensor
nodes deployment
21
Simulation Results Blue superiority
Plot 2a. 5,000 red and 15,000 blue sensor
nodes deployment
22
Deployment Statistics
- field 10 x 10 square units - 20, 000
sensors Statistics at the deployment moment
23
Plot 2a Explanation

• field 10x10 units2 20,000 sensors
• transmission range 1 unit
• 16,700 of sensors sleep gt 3,300 nodes are
  awake gt 3,300 listen or transmit
• Random deployment gt assume these 3,300 awake
  sensors are also distributed randomly and roughly
  uniformly gt on field 10x10 units2 we have
  3,300 awake sensors, meaning 33 awake sensors
  per 1 unit2
• area of transmission region is piTR2 3.14
  units2
• gt 3.14 33 104 sensors are within the
  transmission range of one sensor
• (provided the deployment distribution is uniform).

24
Plot 2a Explanation (2)

• these 104 nodes are awake, meaning some of them
  are
• listening to Reds
• listening to Blues
• transmitting Reds
• transmitting Blues
• its not so critical to determine exactly how
  many transmitting sensors of each color we have,
  but its obvious that there are too many of them
  to get a clear transmission gt we get too many
  collisions

25
Collisions why is it a problem?

• why plot is growing?
• initially 15,000 red nodes and 5,000 blue sensor
  nodes
• at deployment moment

26
Collisions why is it a problem? (2)

• 270 transmitting blue sensors per 100 units2
  communication range of 3.14 units2 ? 8
  transmitting blue sensors per each sensors
  communication range
• 809 transmitting red sensors per 100 units2
  communication range of 3.14 units2 ? 25
  transmitting red sensors

27
How to solve too many collisions issue?

• to reduce the number of collisions each frequency
  
• - red frequency
• - and blue frequency
• is partitioned into 30 channels

28
Collisions Problem, Channels

• to transmit - each sensor randomly picks one of
  30 channels
• to listen to neighborhood - each sensor scans
  through all 30 channels of a frequency
• if two transmitting nodes meet on the same
  channel collision is detected

29
Simulation with 30 channels
Plot 3a. 15,000 red and 5,000 blue sensors
deployment, 30 channels
30
Simulation with 30 channels (2)
Plot 3b. 5,000 red and 15,000 blue sensors
deployment, 30 channels
31
Simulation with 30 channels (2)
Plot 2a. 5,000 red and 15,000 blue sensors,
single frequency
Plot 3b. 5,000 red and 15,000 blue sensors,
30 channels
32
Simulation with 15 channels
Plot 4a. 15,000 red and 5,000 blue sensors
deployment, 15 channels
33
Simulation with 45 channels
Plot 3a. 15,000 red and 5,000 blue sensors,
30 channels
Plot 4b. 15,000 red and 5,000 blue sensors,
45 channels
34
30 channels utilization
35
30 channels Count Collisions (as two)
Plot 3a. 15,000 red and 5,000 blue sensors,
30 channels, Ignore Collisions
Plot 5. 15,000 red and 5,000 blue sensors,
30 channels, Count Collisions (as 2)
36
Individual sensor conservativity importance
Plot 3a. 15,000 red and 5,000 blue sensors,
30 channels, Conservativity 0.25
Plot 5. 15,000 red and 5,000 blue sensors,
30 channels, Conservativity 0.10
37
Individual sensor conservativity importance (2)
Plot 3a. 15,000 red and 5,000 blue sensors,
30 channels, Conservativity 0.25
Plot 5. 15,000 red and 5,000 blue sensors,
30 channels, Conservativity 0
38
Sensors with Simple Memory

• The memory mechanism for this experiment was
  chosen as follows
• sensor gets a memory storage consisting of a
  set of 5 slots
• this set of slots is used for remembering 5
  last colors of the sensor
• where t0 is a current moment in time.

39
Sensors with Simple Memory (2)
Plot 7. 5,000 red and 15,000 blue sensors,
30 channels, sensors with memory
Plot 6. 5,000 red and 15,000 blue sensors,
30 channels, Count Collisions (as 2)
40
Conclusions

• massively deployed sensor network population of
  entities
• emerging behavior
• collisions problem solution
• channels utilization
• useful application of noise collision
• conservativity
• local stabilization leads to the global stability!

41
Future Work

• This project was considered as a first step
  toward solving a real world problem of clock
  synchronization in wireless sensor networks.
• It would be quite interesting to see if using the
  approach and methods of this project one can
  achieve local synchronization in WSN and even
  more challenging if using such locally
  synchronized neighborhoods of sensors one can
  solve WSN global synchronization issue.

42
Acknowledgments

• I would like to thank Dr. Olariu
• for the idea of this project, for his willingness
  to help with all kinds of issues, not punishing
  me for my mistakes and encouraging me to discover
  new layers and to search for new methods and
  solutions
• I also would like to thank my family
• my husband for helping me in my struggle with the
  visual interface for this project and for being
  my editor for the final version of the report
• my mom for keeping my son busy with books and
  drawings
• and my son for interest in my plots, copying them
  but also keeping them intact.

43
References

• K. Jones, K. Lodding, S. Olariu, A. Wadaa, L.
  Wilson, M. Eltoweissy, Biomimetic Models for
  Wireless Sensor Networks
• E. Bonabeau, M. Dorigo, G. Theraulaz,
  Inspiration for Optimization from Social Insect
  Behaviour Nature, Vol. 406, July 6, 2000, pg.
  39
• J. Packard, in press, Social Behavior of Wolves
  Reproduction and Development in Family Groups
  in L.D. Mech and L. Boitani, eds, The Ecology and
  Behavior of the Wolf, The University of Chicago
  Press, Chicago.
• 4. D. Gracanin, M. Eltoweissy, S. Olariu, A.
  Wadaa, Extensible Service-Centric Model for
  Wireless Sensor Networks
• J. Dornstetter, D. Krob, M. Morvan, L. Viennot,
  Some Algorithms for Synchronizing Clocks of Base
  Transceiver Stations in a Cellular Network
  Journal of Parallel and Distributed Computing,
  1999
• 6. K. Kelly, Out of Control The New Biology of
  Machines, Social Systems and the Economic World
  Perseus Books Group, 1995
• 7. Joshua M. Epstein, Robert L. Axtell, Growing
  Artificial Societies Social Science from the
  Bottom Up The MIT Press, 1996

44

• Thank you