ECMANSI - Energy Conserving Multicast for Ad- Hoc Networks with Swam Intelligence Chaiporn Jaikaeo Vinay Sridhara Chien-Chung Shen.

ECMANSI-

Energy Conserving Multicast for Ad-Hoc Networks with Swam

Intelligence

ECMANSI-

Energy Conserving Multicast for Ad-Hoc Networks with Swam

Intelligence

Chaiporn Jaikaeo

Vinay Sridhara

Chien-Chung Shen

• Motivation• Swarm Intelligence – Biological Metaphor• MANSI – Overview• MANSI – Framework Protocol Description• ECMANSI – Description • ECMANSI – Simulation Scenario• ODMRP – Overview • ECMANSI – Simulation Results• Summary and Future work

• More than often Point to Multipoint communication is required

• Power efficient multicast for MANETs– The mobile nodes are energy constrained– The communication module often accounts for a

considerable amount of energy expenditure– Also the topology of the Ad-Hoc networks can be controlled

by adjusting the transmission power of each individual node to reduce interference

– Controlling the transmission power of node may increase the overall network lifetime

MOTIVATION

• Motivation• Swarm Intelligence – Biological Metaphor• MANSI – Overview• MANSI – Framework Protocol Description• ECMANSI – Description • ECMANSI – Simulation Scenario• ODMRP – Overview • ECMANSI – Simulation Results• Summary and Future work

Swarm Intelligence Biological Metaphor

• Essence of Swarm Intelligence– Positive and negative feedback

• search good solutions and stabilize the results

– Amplification of fluctuation• discover new solutions and adapt to changing environment

– Multiple interactions• Allows collaborations among distributed entities to

coordinate and self-organize

A distributed adaptive control system

•Ants likely choose paths with higher pheromone intensity

•Trail gets reinforced(positive feedback)

skip

Ants lay pheromone

Without reinforcement,

pheromone evaporates

(negative feedback)

skip

Most ants follow trail with highest intensity

Most ants follow trail with highest intensity

But some may choose alternate paths with small probability(amplification of fluctuation)

But some may choose alternate paths with small probability(amplification of fluctuation)

Pheromone Trail

• Motivation• Swarm Intelligence – Biological Metaphor• MANSI – Overview• MANSI – Framework Protocol Description• ECMANSI – Description • ECMANSI – Simulation Scenario• ODMRP – Overview • ECMANSI – Simulation Results• Summary and Future work

MANSI – Overview

• Connected subgraph containing all the forwarding nodes are extracted – Forwarding Set

• Forwarding Set consists of nodes which are shared by all the group members – Group Shared Approach

• Forwarding Nodes always rebroadcasts the packet regardless of the previous hop node – Mesh Based Approach

• Forwarding set is constructed only when a node (Source) has data to send – Reactive Approach

MANSI – Multicast for Ad Hoc Networks with Swarm Intelligence MANSI – Multicast for Ad Hoc Networks with Swarm Intelligence

m1

m2

m3

skip

MANSI – Overview

• Each member establishes connectivity to a designed member which serves as the focal point.

• Multicast connectivity is more efficient when group members share existing forwarding nodes

Forwarding set of 6 nodes

m1m3

m2

Forwarding set of 4 nodes

m1m3

m2

• Motivation• Swarm Intelligence – Biological Metaphor• MANSI – Overview• MANSI – Framework Protocol Description• ECMANSI – Description • ECMANSI – Simulation Scenario• ODMRP – Overview • ECMANSI – Simulation Results• Summary and Future work

MANSI-Framework Protocol Operations

Three step process

Source Node

• Neighbor Discovery

– A node which has data to send becomes CORE node and broadcasts a JOIN_REQUEST

– All the nodes that receive the JOIN_REQUEST rebroadcast the request and the one hop neighbor information is discovered

skip

MANSI-Framework Protocol Operations

• Forwarding Set Initialization– A forwarding set is rapidly constructed on-demand– Non-duplicate announcement is rebroadcast by all nodes– Other members request to join the group via the reverse

paths– Requested nodes become forwarding nodes and form

the forwarding set

Core

Core floodsannouncement

Members replyvia reverse paths

Initial Forwarding sethas formed

skip

MANSI-Framework Protocol Operations

• Forwarding Set Evolution– Forward Ants (packets) are deployed by members to

opportunistically learn new connectivity that yields lower cost– A Forward Ant turns into a Backward Ant when it encounters

another existing path and returns to its originator

Core Cost = 3

Cost = 2

Ants follow current best paths and update costs

Core

Ants opportunistically discoverother paths

Cost = 2Cost = 1Total cost = 5 Total cost = 4

skip

MANSI-Framework Protocol Operations

• Pheromone Updating– The Backward Ant deposits pheromone on its return trip to the

sender– The Amount of Pheromone deposited is inversely proportional to

the cost of the trip.• Shorter the trip Higher the pheromone deposition

skip

Core

Cost = 0

Cost = 1Cost = 2m2

m1

turn intoBackward

AntB

C

AD

E

• Motivation• Swarm Intelligence – Biological Metaphor• MANSI – Overview• MANSI – Framework Protocol Description• ECMANSI – Description • ECMANSI – Simulation Scenario• ODMRP – Overview • ECMANSI – Simulation Results• Summary and Future work

Energy Conserving MANSI

• Assumptions– Nodes are capable of adjusting their transmission

power levels (0, Pmax)

– Ant packets and Hello packets are always transmitted at full power

• Objective– Reduce total energy consumed per multicast packet– The cost of a node becoming a forwarding node is the

transmission power level required by the node to transmit the multicast packet

30 mW

20 mW2 mW

2 mW

2 mW

4 mW

8 mW

Total of - 2 transmissions - 50 mW

Total of - 2 transmissions - 50 mW

Total of - 5 transmissions - 18 mW

Total of - 5 transmissions - 18 mW

Energy Conserving MANSI

Number of transmission VS total power consumptionNumber of transmission VS total power consumption

Energy Conserving MANSIRequired Transmission Power

d

fullTXCfullRX pow

pow Generic Propagation Formula

Required Sensitivity powthresh RXRX

pow

powthreshpow

threshpow

powthresh

fullRX

fullTXRXreqTX

C

dRXreqTX

d

reqTXCRX

Required Transmission power

Calculated by transmitting HELLO packet at full powerCalculated by transmitting HELLO packet at full power

Energy Conserving MANSIRequired Transmission Power

Calculated by transmitting HELLO packet at full powerCalculated by transmitting HELLO packet at full power

For data packets, a forwarding node i computes its transmission power level as follows:

where Di is a set of nodes who request and are requested by i to be a forwarding node

ijDj

i reqTxtxPoweri

max ijDj

i reqTxtxPoweri

max

i

i’s communication rangewhen broadcasting data packets

Cost of a node can increase when a farther node choose the current node as a forwarder

Cost of a node can increase when a farther node choose the current node as a forwarder

Energy Conserving MANSI

Join Request Path

ECMANSI – MANSI (Comparative illustration)ECMANSI – MANSI (Comparative illustration)

Energy Conserving MANSI

MANSI

EC-MANSI

Forwarding cost over timeForwarding cost over time

Energy Conserving MANSI

• With mobility, multicast connectivity becomes fragile• (EC)MANSI with mobility-adaptive mechanism

– Each node keeps track of link failure frequency which indicates stability of its surrounding area.

– When link failure frequency is higher than the threshold, a forwarding/member node picks two forwarding nodes with highest pheromone intensities, instead of one

Energy Conserving MANSI

Without mobility-adaptive With mobility-adaptive – more robust group connectivity

Adaptability to mobilityAdaptability to mobility

• Motivation• Swarm Intelligence – Biological Metaphor• MANSI – Overview• MANSI – Framework Protocol Description• ECMANSI – Description • ECMANSI – Simulation Scenario• ODMRP – Overview • ECMANSI – Simulation Results• Summary and Future work

Energy Conserving MANSI

Terrain dimension 1000×1000 m2

# Nodes 50-200

Communication range 250 m

Mobility speed 0-10 m/s

# Members 10

# Senders 1-8

Application Traffic CBR (4512B/s from each sender)

• Ten random networks on QualNet simulator• Multicast sessions running for 30 minutes• Network parameters

Simulation ScenarioSimulation Scenario

• Motivation• Swarm Intelligence – Biological Metaphor• MANSI – Overview• MANSI – Framework Protocol Description• ECMANSI – Description • ECMANSI – Simulation Scenario• ODMRP – Overview • ECMANSI – Simulation Results• Summary and Future work

ODMRP Overview

• On Demand Multicast Routing Protocol(Lee, Su, Gerla – UCLA)

• Each sender floods JOIN-DATA (data+query) to find members on-demand

• Members propagate JOIN-REPLY back to the sender, resulting in mesh creation

• As long as a sender still has data to send, it periodically floods JOIN-DATA to refresh the mesh

• Flooding by every sender causes a scalability problem in large networks, especially with many senders

ODMRP Overview

F

A

B

ED

G

H

I

SenderNextHop

H C

SenderNextHop

H D

SenderNextHop

H HC

skip

• Motivation• Swarm Intelligence – Biological Metaphor• MANSI – Overview• MANSI – Framework Protocol Description• ECMANSI – Description • ECMANSI – Simulation Scenario• ODMRP – Overview • ECMANSI – Simulation Results• Summary and Future work

Energy Conserving MANSI Simulation Results

Network SizeNetwork Size Number of SourcesNumber of Sources MobilityMobility

Packet Delivery RatioPacket Delivery Ratio

Energy consumed per multicast PacketEnergy consumed per multicast Packet

Energy Conserving MANSI Simulation Results

Network SizeNetwork Size Number of SourcesNumber of Sources MobilityMobility

Energy Conserving MANSI Simulation Results

Control Packet OverheadControl Packet Overhead

Network SizeNetwork Size Number of SourcesNumber of Sources MobilityMobility

• Motivation• Swarm Intelligence – Biological Metaphor• MANSI – Overview• MANSI – Framework Protocol Description• ECMANSI – Description • ECMANSI – Simulation Scenario• ODMRP – Overview • ECMANSI – Simulation Results• Summary and Future work

Summary and Future work

• An Energy Conserving Multicast protocol – ECMANSI is proposed

• By adopting the Swarm Intelligence metaphor, forwarding nodes are chosen and their transmission power levels are adjusted dynamically

• Simulation experiments show that energy consumption can be reduced drastically

• Future Work– Incorporate changing of reception power into pheromone

table maintenance, to deal with mobility more appropriately

– Having an extra cost field for each node – amounting to remaining battery power

Questions

Thank You !!!