Energy-Efficient Broadcast and Multicast Trees for Reliable Wireless Communication

Suman Banerjee, Archan Misra, Jihwang Yeo and Ashok Agrawala

IEEE Wireless Communications and Networking Conference

(WCNC) 2003

Speaker: Ju-Mei Li

Outline

Introduction Calculating Energy Costs Retransmission-Aware Minimum

Energy Trees Performance Comparisons Conclusions

Introduction

In wireless environments Individual links often have high error

rates Currently minimum-energy tree

formation algorithms Assign costs to links based on the

energy spent in a signal transmission Do not consider the link error rates

Introduction

Wieseltheir et al [10] Have constant approximation ratios to th

e optimal solution for error-free wireless links

Three broadcast tree formation algorithms

Broadcast Incremental Power (BIP) Broadcast Least-Unicast-cost (BLU) Broadcast Link-based MST (BLiMST)

Introduction: BIP

1

4

3

9

7

6

10

2

8

5

Introduction: BLU and BLiMST

1

4

3

9

7

6

10

2

8

5

1

4

3

9

7

6

10

2

8

5

BLU BLiMST

Assumption

Error rate pi,j for any link (i, j): 30~40% No mobility Packet will be retransmission until all children rec

eive it ACK

Chooses minimum power to reach each child which does not receive broadcast packet

NAK: just one NAK will be received by sender Use a transmission power to reach all children on the tree

Calculating Energy Costs

Packet error rate, S[i].error = p = 1 – (1 - pb)s

S[i].not_rcv_prob = S[i].not_rcv_prob * (S[i].error)tx

pb is the bit error rate and

s is the packet size

Nerfcpb

rconstant.P

Calculating Energy Costs: ComputeTxCost(x, S)

x

i1

i2

i3

i1

The set within the transmission power and is sorted by decreasing order of distance

from x

For i1

Px,i1: transmission a packet from x to i1

Ex,i1: single transmission power

Use Px,i1 i1.error, i2.error and i3.error

Use i1.error tx (number of reliable transmissions of i1)

Use tx, Ex,i1 and i1.not_rcv_prob cost of i1

Update i2.not_rcv_prob and i3.not_rcv_prob

For i2

Px,i2: transmission a packet from x to i1

Ex,i2: single transmission power

Use Px,i2 i1.error, i2.error and i3.error

Use i2.error tx (number of reliable transmissions of i2)

Use tx, Ex,i2 and i2.not_rcv_prob cost of i2

Update i3.not_rcv_prob

i2

i3

For i3

Px,i3: transmission a packet from x to i1

Ex,i3: single transmission power

Use Px,i3 i1.error, i2.error and i3.error

Use i3.error tx (number of reliable transmissions of i3)

Use tx, Ex,i3 and i3.not_rcv_prob cost of i3

Retransmission-Aware Minimum Energy Trees

RBIP RBLU RBLiMST Sweep algorithm Multicast trees

Retransmission-Aware Minimum Energy Trees: RBIP

61

4

3

2

5

7

6

4

3

2

5

7

Update cost and parent of node 5 and

node 6

Update Cost and parent of node 3 and

node 4

Retransmission-Aware Minimum Energy Trees: RBIP

61

4

3

2

5

7

Like the BLU, but the cost of link (i, j) Ei,j (reliable) = Ei,j * (1/1 – pi, j)

Retransmission-Aware Minimum Energy Trees: RBLiMST

Like extension of BLiMST algorithm

61

4

3

2

5

7

Retransmission-Aware Minimum Energy Trees

Sweep algorithm in a post order traversal of the tree Δy = ComputeTxCost(y, Cy – {x}) – ComputeTx

Cost(y, Cy)

Δz = ComputeTxCost(z, Cz {x}) – ComputeTxCost(z, Cz)

x

y

z

Cy

Cy

Cz

Cz

Cz

Retransmission-Aware Minimum Energy Trees

Multicast trees Compute the broadcast tree without

using sweep algorithm Delete nodes which do not lead to

any multicast group numbers (in a single post-order traversal)

Sweep algorithm are performed on the remaining tree (in post-order)

Performance Comparisons

100 nodes Network environment

Random Grid: 100*100 square gird

100 runs for each result

Performance Comparisons: ACK

Performance Comparisons: ACK

Performance Comparisons: ACK

Performance Comparisons: NAK

Performance Comparisons: NAK

Performance Comparisons: NAK

Conclusions

for multi-hop wireless environments Present energy-efficient reliable

broadcast and multicast schemes ACK NAK

End