Performance Evaluation of LEACH Routing Algorithm in Wireless Mobile Sensor Networks

1
Performance Evaluation of LEACH Routing Algorithm
in Wireless Mobile Sensor Networks

• Arash Tavakkol,
• Advanced Network Course Project
• Computer Engineering Department,
• Sharif University of Technology,

2
Outline

• Introduction
• A Quick Safari Through Wireless Sensor Networks
  (WSNs)
• The Important Role of Mobility in WSNs
• Xmulator ( Novel Object Oriented Based Framework
  for Modeling WSNs)
• Simulation Preparation and Results
• Proposed Solution (New Versions of LEACH Routing
  Algorithm)
• Future Works and Conclusions

3
Introduction

• Recent advances in technology enable us to use
  small sensor nodes for environment monitoring.
• Example Smart Building, Smart Hospitals,
  Disastrous Environment Monitoring, Factory
  Monitoring, Battle Fields.
• The available techniques for current networks do
  not support this kind of networks.
• Designing protocol stack for this kind of new
  network is a challanging problem for engineers
  and researchers.
• We can use some of the Ad Hoc Network techniques
  for the design of sensor networks but they differ
  from each other according to the following facts

4

• Difference to Ad Hoc Networks
• Number and density of sensor nodes.
• Sensor nodes are prone to failure.
• Topology of the network is changing rapidly
  (because of mobility, lose of power and ).
• Limit in power of sensor nodes.
• Sensor nodes usually dont have global ID.
• So we must search for new techniques of the
  design.

5
A Quick Safari Through Wireless Sensor Network
System (WSN)
6
Wireless Sensor Network Architecture
7
Designing Factor and Metrics

1. Fault Tolerant The failure of sensor nodes must
   not affect heavily the network functionality.
2. Scalability We can easily change the size of
   network.
3. Low Power We must extend the life time of the
   system.
4. Mobility Our design must support mobile nodes.
5. Task Management We must efficiently distribute
   task among the sensors, so efficient distributed
   algorithms.
6. Product Cost Each sensor must be sufficiently
   low cost.

8
Designing sensor network protocol stack according
to some of it performance metrics.

• In each layer of sensor network we face
  optimization problem according to the described
  metrics.

9
Network Layer

• Usually the power spent for information
  transmition is higher than computational power
  (Pcomp ltlt PTran).
• The model for path loss are Free-Space and
  Multipath Fading.
• The key point in wireless sensor networks is
  Multihop routing.

10
Classification of Routing Algorithms for WSNs
11
Low Energy Adaptive Clustering Hierarchy (LEACH)
12
Low Energy Adaptive Clustering Hierarchy (LEACH)

• In this approach WSN is divided into clusters.
• Each cluster collects and compresses the data
  form cluster members and sends to base station.
• Each cluster member sends its data to clusterhead
  periodically according to TDMA table.
• After specific time, the configuration of the WSN
  will be changed and new cluster head will be
  selected.
• LEACH algorithm has the following 3 phases

13

• Cluster Head Selection Phase
• The aim is to have in average K clusterhead in
  each round.
• Suppose we are in roundr1.
• Each node generates uniform random variable R.
• If RltT(n) then the node presents itself as
  cluster head.
• The threshold is computed as follow
• In this equation G is the set of nodes that are
  not selected as clusterhead in recent N/K round.

14
LEACH Algorithms
15

• Setup Cluster Phase
• Each cluster head broadcasts ClusterHeadID
  message with its ID in network (CSMA).
• Each non cluster head listens to these messages
  and selects appropriate cluster head with respect
  to its received signal energy.
• Selecting appropriate cluster head, each node
  broadcast its ID and ID of cluster head with
  message JIONTOCH in network (CSMA).
• After each clusterhead finds its members, it
  makes TDMA table for data transmission for each
  member and send to them.
• Each cluster member finds its time and turns off
  its radio till its turn and sends data.
• After each time slot each clusterhead compresses
  data and sends it to base station.

16

• After certain period of time the clusterhead
  changes and new nodes become cluster head.

LEACH At the end of steady phase
17
LEACH in one Glance
18
Mobility

• One of the most critical characteristics of WSNs
  is Mobility.
• Concurrent usage of Wireless Network and Mobile
  Network in the literature illustrates the
  importance of this attribute.
• Mobility affects different layers like MAC and
  Network and usually leads to complexity in the
  algorithms
• For example the more the nodes change their
  position the more routing tables become invalid.

19
Research Areas About Mobility

• Physical studies Investigating nodes
  displacement from physical point of view and
  without any consideration to signaling effects.
• Abstract modeling of nodes displacement.
• Very important for simulation studies
• Middle Performance Models Analytical modeling of
  the effect of mobility on the routing protocols
  and topology control protocols.
• Mobility in Topology Control and Routing
  Algorithm Investigating the effect of mobility
  on different layers of WSNs and proposing
  solutions to prevent performance degradation.

20
Physical Studies

• Proposes an Analytical model for mobility.
• The model should have following properties to be
  useful
• Having enough simplicity to be implemented in
  simulation studies
• Does not have great difference to real values.
• Currently two types of models used in
  simulations
• Trace based models In this method realistic
  mobility patterns used for simulation
• Analytical models

21
Analytical Mobility Models

• Random Walk Model Each node changes its current
  position with choosing two parameters Velocity
  and degree of displacement.
• The movement continues till a predefined time or
  distance.
• At the end of movement a new velocity and degree
  will be selected for new movement.
• This models has no memory which leads to
  instantaneous stop or great direction changes.

22
Analytical Mobility Models (contd)

• Random Way Point In this model we have Pause
  Time, which means a node can have certain periods
  without any movement.
• Each nodes waits for a specific period and at the
  end of this period selects a destination and
  moves toward it.
• The displacement speed is chose from a limited
  interval.
• After arrival this process will be repeated.
• This model has been used in many academic
  researches and we have used it in our
  investigation for LEACH.
• It is very simple but the results will be stabled
  very late. So the simulation time will be too
  long.

23
Analytical Mobility Models (contd)

• Group Mobility Models Used for modeling the
  mobility of a group of objects. The most famous
  models in this category are
• Exponential Correlated Random Mobility Model A
  probabilistic function used for all nodes to give
  correlation.
• Column Mobility Model Used for displacements
  like a column of soldiers. Te members of the
  column move in a same direction with same speed.
• Nomadic Community Mobility Model
• Reference Point Group Mobility Model

24
Other Physical Models

• Random direction model Provided uniform
  distribution of nodes.
• Boundless simulation area solving the problem of
  instantaneous change in speed and direction.
• Gauss-Markov model
• City Section Mobility Model
• Manhattan Mobility Model
• Highway Mobility Model

25
Xmulator ( Novel Object Oriented Based Framework
for Modeling WSNs)

• Motivation for Developing a Modeling Platform for
  WSNs
• Possibility of defining details of algorithms
  with performance considerations.
• There is a lack of good simulators( J-Sim, NS,
  OpNet).
• Using new software architecture paradigms (OOD).
• Using OO programming languages (C , Java).
• XmulatorXMLSimulator.
• Using XML format for defining topologies,
  parameters, and outputs.
• Ease of developing new packages, new tools for
  existing packages.
• Possibility of debugging even into the core of
  simulator.
• Logging capabilities.

26
Principles of Simulation

• Time slice Method
• In this method we divide the time to small time
  slices and run the program (Master Clock)
• Small time slices
  reduction in performance
• Large time slices
  reduction the accuracy
• Process based Method
• In this method each part of the real system is
  modeled with a process.
• easy of defining the behaviors of the components.
• low performance.
• Event based Method
• Record the events to simulate the system
  behavior.
• Xmulator using this method in this version.

27
Simulation Preparation

• We have simulated under two size for networks
• 200 nodes
• 500 nodes
• 1000 nodes
• 2000 nodes
• Maximum Transmission Power 0.6 10-3 Wat.
• Path Loss 1 / d2
• Listening Threshold 10-8 Wat.
• Uniform distribution of nodes along each axes.
• Percent of Cluster Nodes 10 , 20
• Initial Platform Size 400 400 meter to
  10001000 meter
• Message Length 1024 bit

28

• Simulation Results
• (Throughput)

29
(No Transcript)
30
(No Transcript)
31

• Simulation Results
• (Energy)

32
(No Transcript)
33
Conclusions And Future Works

• Mobility is one of the most important properties
  of Wireless Sensor Networks and should be
  considered in design of each layer of network.
• The simulation results depict that LEACH routing
  algorithm is not mobility tolerant and its
  performance is highly sensitive to mobility.
• The Exponential Correlated Random Mobility Model
  has less degradation effect on the performance of
  LEACH in comparison to Random Way Point Mobility
  Model.
• The energy dissipation of LEACH grows too much
  under the random way point mobility model. This
  is another fact that shows this algorithm is not
  mobility tolerant.

34

• Future Work
• Investigating the effect of other mobility models
  on LEACH.
• Comparison of LEACH to other routing algorithms
  from mobility point of view.
• Proposing solutions/revisions for LEACH to make
  it mobility tolerant.
• THE END