Energy-Efficient Shortest Path Self-Stabilizing Multicast

Protocol for Mobile Ad Hoc Networks

Ganesh SridharanGanesh.Sridharan@asu.edu

Outline

Introduction Goals System Model Cost metric Simulation & Implementation Conclusion

Introduction

Mobile Ad Hoc Networks (MANETs) No infrastructure Limited transmission range Energy constrained

Multicasting in MANETs Why multicast as opposed to multiple

unicast? Less number of messages Less energy spent

Introduction

Issues in MANET Multicasting Dynamic Topology Energy constrained Possible solution – flooding

Suffers from redundant rebroadcast Increase in collision and contention Energy inefficient

Tree or Mesh Structure Examples: MAODV, ODMRP etc.

Shortest Path Self-Stabilizing ProtocolSS-SPST

Shortest path spanning tree from root Pro active tree construction Tree includes both multicast group and non-group nodes

Faults Change in topology caused by mobility

SS-SPST is self-stabilizing Converge to a global legitimate state from an illegitimate

state Fault-tolerant solution

SS-SPST is distributed Uses only local knowledge

Self-Stabilization Properties

Convergence Closure

Inter-communication Share memory Message passing

Beaconing Time complexity

Rounds Round definition in a lossy medium

A round is defined to be the time period in which each node in the system receives at least one beacon message from each of its neighbors and performs computation based on the information it has received.

SS-SPST Cost metric

Multicast tree is constructed to optimize the cost metric

Currently hop count is the cost metric

Goal: To optimize energy An energy-efficient cost metric is required

to minimize total energy consumption

Wireless Multicast Advantage

X

Y

Z

PXZ

PXY

PXZ > PXY

Motivation - example

R

1 2

NG

NG

NG

X

Total discard energy = 3 * Reception energy

Problem Statement

Propose energy-efficient cost metric Simulation based performance

comparison with MAODV and ODMRP Comparison of different cost metrics

MAODV & ODMRP

MAODV properties Tree based On-demand Route request and route reply phase

ODMRP properties Mesh based On-demand Many routes to the receivers

System Model - Assumptions

Unique identification Periodic beaconing Soft-state neighbors Cost metric computation Dynamic transmission range Active mode Single source multicasting

Energy Model

ETx = Eelec . K + Eamp . K . d2

ERx = Eelec . K

Eelec = Fixed energy

Eamp = Amplification energy

K = Number of bits d = distance

SS-SPST - Algorithm If (root)

Dist-to-root = 0Parent = -1

elseDist-to-root = Shortest

distance to root through any

neighbor node ‘i’

Parent = i

R

1 2

NG

NG

NG

X

SS-SPST An Example

SS-SPST An Example

R

1 2

NG

XNG

NG

Round 1

Round 2

Round 3

Motivation - example

R

1 2

NG

NG

NG

X

Total discard energy = 3 * Reception energy

Cost metric Hop count

Cij = 1 Transmission Energy

Cij = Tij

Transmission Energy based on farthest node Cij = (Tij + R) if j is the farthest

node from i = R otherwise

Cost metric

Transmission Energy based on farthest node with discard energy

Cij = (Tij+R+Li) if j is the farthest node from i

= R otherwise

Li = R * (#neighborsi - #tree childreni)

An Example

0

1

6 2

345

7

8 9

120.1

120.02

75.27

75.37

120.36

120.04

120.56

120.06

200.03

120.45 120.34

75.48 75.49

Hop count metric – SS-SPST

Stabilization time = 3 rounds

Energy consumed / bit = 5.95 micro J

0

1

6 2

345

7

8 9

1

1

1

1

1

1

1

1 1

Round 1

Round 2

Round 3

Transmission Energy metric – SS-SPST-T

Stabilization time = 4 rounds

Energy consumed / bit = 4.72 micro J

0

1

6 2

345

7

8 9

1.492

1.49

0.617

1.4909

1.491

4.051

1.5

0.619 0.6199

Round 1

Round 2

Round 3

0.618Round 4

Max Transmission Energy metric – SS-SPST-F

Stabilization time = 5 rounds

Energy consumed / bit = 3.392 micro J

0

1

6 2

345

7

8 9

0.05

0.05

0.617

0.05

0.05

4.101

1.55

0.05 0.05

Round 1

Round 2

Round 3

Round 40.05

Round 5

1.542

Max Transmission Energy + Discard Energy metric – SS-SPST-E

Stabilization time = 5 rounds

Energy consumed / bit = 3.29 micro J

0

1

6 2

345

7

8 9

0.05

0.05

0.657

0.05

0.05

4.101

1.55

0.05 0.05

Round 1

Round 2

Round 3

Round 4

0.05

Round 5

1.542

Summary

Metric # rounds Energy in micro J

SS-SPST 3 5.9512

SS-SPST-T 4 4.7279

SS-SPST-F 5 3.3922

SS-SPST-E 5 3.2959

Simulation Environment

Simulator - NS-2 Simulation area - 750 x 750 Simulation time - 1800 seconds # nodes - 50 Traffic rate – 64 Kbps # group nodes - 20

Performance Metrics

Packet delivery ratio #pkts received/#pkts transmitted

Energy consumed per packet delivered Total energy consumption/pkts

received End-to-end delay

Total delay per packet/#received nodes

Unavailability ratio Service interrupt time/simulation

time

Energy Spent – Different cost metrics

Packet Delivery Ratio – Different cost metrics

Unavailability Ratio – Different cost metrics

Packet Delivery Ratio – Different protocols

Energy Spent – Different protocols

Control Byte Overhead – Different protocols

Delay – Different protocols

Implementation

To check the correctness of the protocols

Implementation testing with 3 laptops working in ad hoc mode

Emulation – mobility, energy and bit error rate

Implementation

UtilityClasses

PacketListener

Event Handler

SS-SPST

Packet Handler

Send Receive

Conclusion & Future work

Energy saving using proposed cost metric

Cost of saving energy Nodes operating in sleep mode Testing real implementation with

many nodes

Questions?

Thank you!