Performance Evaluation and Improvementof an Ad Hoc Wireless Network

Takayuki YamamotoGraduate School of Engineering Science,

Osaka University, Japantak-ymmt@ics.es.osaka-u.ac.jp

31 Jan 2002 ICOIN-16 2

ContentsIntroduction of Target System - the Flexible Radio Network -Basic Network Performance forSome Different System Parameter ValuesSuggestions for Performance ImprovementConclusion

31 Jan 2002 ICOIN-16 3

Flexible Radio NetworkWireless Ad Hoc Network Systemdeveloped by Fuji Electric Co. Ltd.Multi-hop Network organized byStationary TerminalsApplication Examples

Power Consumption Collectionin Manufacturing PlantsSales Account Collectionfrom Vending MachinesUsage Data Collection from Ski Lift Gates

http://www.fujielectric.co.jp/eng/index.html

31 Jan 2002 ICOIN-16 4

System DescriptionNetwork Configuration Table

updated by Periodic Table ExchangeEach Node maintains Multiple Routesto All Nodes in the Same NetworkRoute Entry = (Neighbor ID, Hop Count)

A

E

B

D

C

‥(B,3)(C,4)2

：：：

‥(C,2)(B,1)1

‥EB

Destination Node ID

Route

entr

ies

Network Configuration Table at node A

31 Jan 2002 ICOIN-16 5

Protocol DescriptionTime Division, Slot-based Transmission

Packet Maximum Lifetime set in a Source Nodedecreased by one per slot even when it stays in a buffer

Hop-by-Hop Receipt Acknowledgementbased on Relay Echo Mechanism

Slot 0 1 2 3 4 5

Node A

Node B

send

forward(relay)

relay echoNode A eavesdrops node B’sforward transmission.When node A’s first transmissionsuccesses, node A can hear therelay echo and delete the packetin its buffer.

31 Jan 2002 ICOIN-16 6

Protocol Description

Retransmission Control

Slot 0 1 2 3 4

Node A

Node B

transmission failure× retransmission to node B or another node

If no relay echo can be received,node A retransmits the packet.

31 Jan 2002 ICOIN-16 7

Protocol Description

ACK from Destination

Slot 0 1 2 3 4

Node A

Node B

Destination Node returns an ACKpacket by way of relay echo

to B

ACK

31 Jan 2002 ICOIN-16 8

Simulation Description for Original System EvaluationNetwork Simulator - ns-2with our implementation of FRNMaximum Lifetime = 12, 64 and 128 (slots)Data Collection Network Model

31 Jan 2002 ICOIN-16 9

Original System Evaluation

Throughput = Number of Packets received by the Destination per slotPacket Loss Rate = Ratio of unreached Packets to the DestinationLoad = Average Number of Packets produced per slotL(h) = Maximum Lifetime ( h is described later)

Throughput Packet Loss Rate

Long Lifetime Packetcauses Congestion

Short Lifetime Packetcauses High Loss Ratefor Low Load Environment

31 Jan 2002 ICOIN-16 10

Performance Improvement

Adaptive Maximum LifetimeSource Node calculates the Maximum Lifetime for Each Packet according to the shortest Hop Count to its Destination

Methods to Prevent Packet Collision leads to Packet Duplication

Random Wait for RetransmissionPacket Rejection earlier than its Lifetime expires

31 Jan 2002 ICOIN-16 11

Adaptive Maximum Lifetime

A

E

B

D

C F

G

Example Function: L(h) = 6h+12( h is Hop Count of the shortest route)

L(h)=18

L(h)=24

L(h)=30

31 Jan 2002 ICOIN-16 12

Adaptive Maximum Lifetime

Low Loss Rate forLow Network Load

Long-living Packet can berejected effectively

31 Jan 2002 ICOIN-16 13

Packet Duplication Problem

Packet TransmissionFailure

Relay Echo ReceiptFailure

Packet Retransmission

Unique Packetin the Network

PacketDuplication

Additional Network Loadand Performance Degradation

31 Jan 2002 ICOIN-16 14

Random Wait for RetransmissionReduction of Packet Duplication Repeat

Slot 0 1 2 3 4 5

Node X

Node A

Node Bsend

forward

collision×echo loss

send×

echo

resend

forward

×echo loss

resend×collisio

n

echo

Packet Collision repeatsdue to Fixed Wait Duration

Case 1: Fixed Wait Duration (3 slots)

31 Jan 2002 ICOIN-16 15

Random Wait for RetransmissionReduction of Packet Duplication Repeat

Slot 0 1 2 3 4 5

Node X

Node A

Node Bsend

forward

collision×echo loss

send×

echo

resend

forward

echo

resend

echo

Probability of repeatedecho loss can be reducedby random delay time

Case 2: Random Wait Duration

31 Jan 2002 ICOIN-16 16

Random Wait for Retransmission

This Method is effective underLarge Maximum Lifetime Setting

31 Jan 2002 ICOIN-16 17

Early Packet RejectionPacket Rejection earlier than its Lifetime reaches Zero

Unreachable Packets cause Useless Collisions

(X, D, 2)

(source, destination, lifetime)Packet ：

(X, D, 1) (X, D, 0)

Detect and Reject the unreachable Packet whose Lifetime is shorter than the Shortest Hop Count to its Destination

D

31 Jan 2002 ICOIN-16 18

Early Packet Rejection

This Method is effective underSmall Maximum Lifetime Setting

31 Jan 2002 ICOIN-16 19

PLR for All Methods Together

31 Jan 2002 ICOIN-16 20

Conclusion and Future WorkFlexible Radio Network (FRN) system

Relay Echo Mechanism, Multiple Route Maintenance for Reliability

Maximum Lifetime Calculation with Route Lengthis capable of increasing Network Reliability

Packet Duplication caused by Relay Echo and Retransmission Mechanism

Relationship between Maximum Lifetime Setting and Effect of Random Wait Duration, Early Packet RejectionMuch Better Performance is achieved by applying All Methods Together

Future WorksEvaluation for End-to-End Protocol on FRNImplementation into actual FRN Terminals