A Dead-End Free Topology Maintenance Protocol for Geographic Forwarding in Wireless Sensor Networks

IEEE Transactions on Computers, vol. 60, no. 11, November 2011

Chih-Hsun Anthony Chou1, Kuo-Feng Ssu2, Hewijin Christine Jiau2, Wei-Tong Wang2 and Chao Wang2

1Institute for Information Industry, Taiwan2National Cheng Kung University, Taiwan

Outline Introduction Assumptions and Background Dead-End Free Topology Maintenance (DFTM) Protocol Discussions and Analysis Experimental Results Conclusion

Introduction

Topology management schemes have emerged as a promising strategy for prolonging the lifetimes of wireless sensor networks.

Several schemes construct a virtual communication backbone by turning off redundant sensor nodes. a connected dominating set (CDS)

Introduction

The CDS is constructed in such a way Each node is either a member of the subset or is a neighbor of one

of the nodes in the subset.

S

D

Introduction

Dead-End Node Problem

This paper proposes a topology maintenance scheme for the construction of dead-end free topologies in WSNs.

Assumptions and Background

There are many stationary sensors distributed over the monitoring region.

The network is assumed to be sufficiently dense to construct a dead-end free topology.

Each sensor can be in either an active mode or a sleep mode.

Each sensor knows both its own and all its neighbors’ coordinates.

DFTM Scheme Dead-End Free Verification Dead-End Free Topology Construction Dead-End Free Topology Maintenance

Dead-End Free Verification Global Dead-End Free (GDF) Condition

The dead-end situation does not occur at any node in the network

Local Dead-End Free (LDF) Condition

N A

Sleeping neighbors

Active neighbors

Node N does not satisfy the LDF condition.Node N satisfies the LDF condition.

Transmission range

A

Dead-End Free Topology Construction

N

Sleeping neighbors

Active neighbors

Transmission range

Undecided neighbors

B

C

D

E

Active Neighbor Set (ANS)

A B C

Tentative Neighbor Set (TNS)

D E F

F

Node N does not satisfy the LDF condition.

Active Node Selection Algorithm

EA

Dead-End Free Topology Construction

N

Sleeping neighbors

Active neighbors

Transmission range

Undecided neighbors

B

C

D

FActive Neighbor Set (ANS)

A B C F

Tentative Neighbor Set (TNS)

D E

F

Active Node Selection Algorithm

Ruled

The distance between the candidate node and the initiator

AN

B

C

D

F

E

F

Active Node Selection Algorithm

Rules

The length of the new covered segment

A

N

B

C

D

F

E

F

Active Node Selection Algorithm

Preference weighting

r

d

r

ncs aia

1

drule 5.0

i: initiator

r: the node’s transmission range

ncsa: the length of the new covered segment of node a

: the distance from node i to node a aid

srule 5.0ruledrules

Dead-End Free Topology Maintenance Global Topology Maintenance

For energy balancing

All nodes change modes to undecided every Tglobal seconds. The sink node randomly chooses a node to be the initiator. Every node has an equal probability of becoming an active node.

Dead-End Free Topology Maintenance

Local Topology Maintenance Some of the active nodes may suddenly become unavailable. Wake up sleeping nodes

Each active node periodically exchanges messages with its active neighbors.

When a node fails to receive an ACK message from a specific neighbor

The neighbor will be removed from the Active Neighbor Set.

Execute the dead-end free topology construction process.

F

Dead-End Free Topology Maintenance

Local Topology Maintenance Some of the active nodes may suddenly become unavailable.

AN

B

C

D

EE

Discussions and Analysis

Discussions Lemma 1. A network topology is fully connected if it satisfies the

Global Dead-End Free (GDF) condition.

Theorem 1. A network topology constructed by the proposed DFTM scheme is fully connected.

Discussions and Analysis

Analysis The total number of active nodes required in GAF and DFTM.

GAF

Discussions and Analysis

Analysis The total number of active nodes required in GAF and DFTM.

DFTM – Best Case

Discussions and Analysis

Analysis The total number of active nodes required in GAF and DFTM.

DFTM – Worst Case

Discussions and Analysis

Analysis The total number of active nodes required in GAF and DFTM.

d

Experimental Results

ns2 Simulator 50, 75, or 100 static nodes were randomly distributed

within a sensing area measuring 60*30 m. Transmission range of each node: 15 m. Comparisons: GPSR, GAF and SPAN

The length of each GAF square was set to m. SPAN: backbone infrastructure

5/15

Number of Active Nodes50 nodes 75 nodes

100 nodes

Number of Survived Nodes50 nodes 75 nodes

100 nodes

Packet Delivery Ratio50 nodes 75 nodes

100 nodes

Energy Consumption and Path Length

Comparison for Dead-End Occurrence

50 nodes 100 nodes

Conclusion This paper presented a distributed dead-end free topology maintenance

protocol, namely DFTM.

DFTM can be integrated with any geographic routing Low energy consumption A minimum number of dead-end events

The performance of DFTM has been benchmarked against that of GAF and SPAN using the ns2 simulator.

Thank You ~