1 mitigate the bottleneck of underwater acoustic sensor networks via priority scheduling junjie...

27
1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

Upload: john-patrick

Post on 17-Jan-2016

220 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

1

Mitigate the Bottleneck of Underwater Acoustic Sensor

Networks via Priority SchedulingJunjie Xiong, Michael R. Lyu, Kam-Wing

Ng

Page 2: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

Background

Wireless sensor networks (WSNs) Underwater acoustic sensor networks (UWASNs)

Deployed in oceans Difference from terrestrial wireless sensor networks (TWSNs)

Wireless medium Sound (Electromagnetic waves)

Longer latency Higher cost Sparser deployment

2

RegionRegion

Base stationBase station

Sensor nodes

Sensor nodes

Page 3: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

Ocean bottom monitoring by UWASNs

3

Page 4: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

4

Outline Motivation Protocol Design Evaluation Conclusions

Page 5: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

5

Motivation

CTS8ms

CTS8ms

RTS8ms

RTS8ms Propagation time

740msPropagation time

740ms

DATA80ms

DATA80ms

ACK8ms

ACK8ms

CTS8ms

CTS8ms

RTS8ms

RTS8ms

Propagation time 3.7us

Propagation time 3.7us

DATA80ms

DATA80ms

ACK8ms

ACK8ms

SenderSender

ReceiverReceiver TT

A DATA transmission in UWASNs with CSMA/CA

SenderSender

ReceiverReceiver TT

A DATA transmission in TWSNs with CSMA/CA

Page 6: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

Motivation

6

DATA1: P->QDATA1: P->Q T

1T1

I1I1

R1R1

DATA2: Q->SDATA2: Q->S

I2I2

R2R2

T2T2

T3T3

R3R3

DATA3: S->QDATA3: S->Q

I3I3

DATA4: P->QDATA4: P->Q T

4T4

I4I4

R4R4

T stands for transmittingR stands for receivingI stands for interference

T stands for transmittingR stands for receivingI stands for interference

Node PNode P

Node QNode Q

Node SNode S

TT

TT

TT

Data transmission between 3 nodes in UWASNs

Data transmission between 3 nodes in UWASNs

DATA4: P->SDATA4: P->S T

4T4

I4I4

R4R4

Collision!

R2

R3

I1

R1

Page 7: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

7

Motivation

T stands for transmittingR stands for receivingI stands for interference

T stands for transmittingR stands for receivingI stands for interference

Node PNode P

Node QNode Q

Base stationBase station

CycleCycle

CycleCycle

CycleCycle

Data transmission between 3 nodesData transmission between 3 nodes

TPTP

IPIP

RPRP

TPTP

IPIP

RPRP

TPTP

IPIP

RPRP

TPTP

IPIP

RPRP

TPTP

IPIP

RPRP

TPTP

IPIP

RPRP

TPTP

IPIP

RPRP

TPTP

IPIP

RPRP

IQIQ

RQRQ

TQTQ

IQIQ

RQRQ

TQTQ

IQIQ

RQRQ

TQTQ

IQIQ

RQRQ

TQTQ

IQIQ

RQRQ

TQTQ

IQIQ

RQRQ

TQTQ

IQIQ

RQRQ

TQTQ

Utilize the propagation time Maximize the throughput by minimizing working period in a cycle

Page 8: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

8

Outline Motivation Protocol Design Evaluation Conclusions

Page 9: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

9

A Routing and Application based Scheduling Protocol (RAS)

DATA1: P->QDATA1: P->Q T

1T1

I1I1

R1R1

T stands for transmittingR stands for receivingI stands for interference

T stands for transmittingR stands for receivingI stands for interference

Node PNode P

Node QNode Q

Node SNode S

TT

TT

TT

Data transmission between 3 nodes in UWASNs

Data transmission between 3 nodes in UWASNs

I1

R1

One time

One time

Three times

Three times

We discover a unique scheduling problem that exists only in UWASNs: different scheduling element.

Page 10: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

10

A Routing and Application based Scheduling Protocol (RAS)

We design the Routing and Application based Scheduling protocol (RAS) which is composed of : The TDMA based MAC mechanism The static routing calculation Centralized schedule calculation by utilizing application data direction

Calculate the traffic of each node. Schedule the traffic transmission and reception.

Page 11: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

11

A Routing and Application based Scheduling Protocol (RAS)

Scheduling principles of RAS At a node, guarantee a DR will not overlap any DT. At a node, guarantee a DR will not overlap any IR. At a node, a DT and one or more IR can coexist. No DR from i-th hop node to (i+1)-th hop node. At a node, use DR as the scheduling basis rather than DT or IR.

DR: data receptionIR: interference receptionDT: data transmission

DR: data receptionIR: interference receptionDT: data transmission

DRDT DT Node m

TDATA

Scheduled DT

Page 12: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

12

To be integrated to the previous slide

DRDT DT Node m

TDATA

DRIR IR Node m

TDATA

Scheduled DT

Scheduled IR

Page 13: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

13

A Routing and Application based Scheduling Protocol (RAS)

Congestion avoidance algorithm Besides throughput, queue length, fairness are important Step1: Schedule the BS's DR from 1 hop nodes. Step2: Schedule the DR tier by tier: from inner tier to outer tier. Step3: For each node m that is going to receive data packets

from its children, arrange its DR from its children alternatively.

Page 14: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

14

Outline Motivation Protocol Design Evaluation Conclusions

Page 15: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

15

Performance Evaluation

Network Throughput

Page 16: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

16

Performance Evaluation

Average End-to-end Delay

Page 17: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

17

Performance Evaluation

Average Maximum Queue Length per Node

Page 18: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

18

Performance Evaluation

Time Slots Required for the RAS Network Working Period

Page 19: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

19

Outline Motivation Protocol Design Evaluation Conclusions

Page 20: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

20

Conclusion

We discover a unique scheduling problem that exists only in UWASNs.

A priority scheduling protocol RAS is designed to provide communications for UWASNs. RAS allows parallel transmissions, and thus improve the

throughput and delay performance based on the scheduling principles.

RAS mitigates queue overflow and is scalable in calculating proper schedules.

Page 21: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

21

Q & AThank you!

Page 22: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

UW-FlASHR

UW-FLASHR (Achieving high channel utilization in a time-based acoustic mac protocol, ACM Mobicom Workshop on UnderWater Networks, 2008) is a distributed TDMA based MAC protocol. utilizes propagation delay to increase throughput. employs no energy-saving mechanism. suffers from collisions.

22

Page 23: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

23

A Cross-layer Communications Protocol Design

Advantages of RAS Reduces mutual communications Reduces energy consumption Avoids collision, increases throughput, and reduces delay and

queue overflow probability for each node

Page 24: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

24

A Cross-layer Communications Protocol Design

Scheduling algorithm formulation Maximize the throughput by minimizing the working period of a cycle

The number of the decision variables equals the number of the packets to be received in a cycle.

Unique scheduling element: the combination of a DR, a DT and a sequence of IR to the other nodes.

The schedule principles.

The time node m receives the Qmjw-th packet from its children node Cmj

Page 25: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

25

A Cross-layer Communications Protocol Design

Congestion avoidance algorithm

Page 26: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

26

A Cross-layer Communications Protocol Design

Congestion avoidance algorithm

Page 27: 1 Mitigate the Bottleneck of Underwater Acoustic Sensor Networks via Priority Scheduling Junjie Xiong, Michael R. Lyu, Kam-Wing Ng

27