query aggregation for providing efficient data services in sensor networks

24
Query Aggregation for Providing Efficient Data Services in Sensor Networks Wei Yu * , Thang Nam Le + , Dong Xuan + and Wei Zh ao * * Computer Science Department Texas A&M University + Department of Computer Science and Engineeri ng The Ohio State University IEEE Mobile Ad-hoc and Sensor Systems (MASS), Shin_wei H

Upload: beau-moore

Post on 02-Jan-2016

18 views

Category:

Documents


0 download

DESCRIPTION

Query Aggregation for Providing Efficient Data Services in Sensor Networks. Wei Yu * , Thang Nam Le + , Dong Xuan + and Wei Zhao * * Computer Science Department Texas A&M University + Department of Computer Science and Engineering The Ohio State University - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Query Aggregation for Providing Efficient Data Services in Sensor Networks

Query Aggregation for Providing Efficient Data Services in Sensor Networks

Wei Yu*, Thang Nam Le+, Dong Xuan

+ and Wei Zhao*

*Computer Science Department

Texas A&M University+Department of Computer Science and Engineering

The Ohio State University

IEEE Mobile Ad-hoc and Sensor Systems (MASS), 2004

Shin_wei Ho

Page 2: Query Aggregation for Providing Efficient Data Services in Sensor Networks

Outline

Introduction Query Aggregation-Based Data

Service Frameworks Weighted Zone-based Query

Aggregation Algorithm Performance Evaluation Conclusion

Page 3: Query Aggregation for Providing Efficient Data Services in Sensor Networks

Introduction

The wireless sensor networks are required to provide efficient data services as a distributed database.

The application can submit its requests as queries.

Page 4: Query Aggregation for Providing Efficient Data Services in Sensor Networks

Introduction (cont’d)

Sensor networks are deployed for monitoring the environment consisting of Temperature sensors Humidity sensors Wind sensors

Such networks typically need to support a large number of users.

Page 5: Query Aggregation for Providing Efficient Data Services in Sensor Networks

Introduction (cont’d)

There are salient features that all of the above application share: query rate can be high the energy consumption spent on sending and

routing queries may far exceed

For these class of applications, optimizing query dissemination is critical to improve performance of the sensor network.

Page 6: Query Aggregation for Providing Efficient Data Services in Sensor Networks

Introduction (cont’d)

In the traditional query dissemination model, applications forward queries to the base station of the sensor networks. processes the queries one by one

This simple approach suffers from shortcomings: Applications may pose duplicate queries Overlapping queries

Page 7: Query Aggregation for Providing Efficient Data Services in Sensor Networks

Introduction (cont’d)

Page 8: Query Aggregation for Providing Efficient Data Services in Sensor Networks

Query Aggregation-Based Data Service Frameworks

Two major problems aggregating the queries routing queries efficiently to proper regions

We discuss three frameworks to solve these problems: Purely Sensor Network-based Framework (PSNF) Purely Base Station-Oriented Framework (PBSOF) Integrated Query Aggregation Framework (IQAF)

Page 9: Query Aggregation for Providing Efficient Data Services in Sensor Networks

Query Aggregation-Based Data Service Frameworks-- Purely sensor network-based framework (PSNF)

Base StationQuery

QueryQuery

Without conducting query aggregation

decision

send the same data multiple times

to reply for different queries

Page 10: Query Aggregation for Providing Efficient Data Services in Sensor Networks

Query Aggregation-Based Data Service Frameworks-- Purely base station-oriented framework (PBSOF)

Base StationQuery

QueryQuery

makes the query aggregation decision based on the input query information.

QueryQuery

New Query

Page 11: Query Aggregation for Providing Efficient Data Services in Sensor Networks

Query Aggregation-Based Data Service Frameworks-- Integrated query aggregation framework (IQAF)

We consider the fact base station has a global picture of all input queries sensor network can take certain roles to execute the

aggregated query plan

Thus, a number of sensor nodes as access nodes are selected as the query proxy.

Page 12: Query Aggregation for Providing Efficient Data Services in Sensor Networks

Query Aggregation-Based Data Service Frameworks-- Integrated query aggregation framework (IQAF) (cont’d)

Page 13: Query Aggregation for Providing Efficient Data Services in Sensor Networks

Weighted Zone-based Query Aggregation Algorithm-- Problem Definition

jEE

QHQGQFE

NttQ

QQQQfunctionnaggregatio

QQQQ

j

i

M

iiii

iii

MN

N

1 ,

function a find tois ovjective The

)]()()([

queriesoutput with for n consumptioenergy The

queries. original )1(by merged is

}, ..., ,{} ..., ,{:

set by denoted ,..., :queries

1

'''

i

'

''

11

1

Page 14: Query Aggregation for Providing Efficient Data Services in Sensor Networks

Weighted Zone-based Query Aggregation Algorithm

Q1(v1)

Q2(v2)

Q3(v3)

Q4(v4)

Q5(v5)

: Query region

Q: Input query

V: Attribute information

Q6

Process the input queries in set Q by filtering queries with full cover property.

Page 15: Query Aggregation for Providing Efficient Data Services in Sensor Networks

Weighted Zone-based Query Aggregation Algorithm (cont’d)

Q1(v1)

Q2(v2)

Q3(v3)

Q4(v4)

Q5(v5)

: Query region

Q: Input query

V: Attribute information

Calculate the overlapping zone and assign the weight

t

ijiijt

iji

OOO

WQQSO

SW

},,{,

2

O1 O2

O3O4 O5

554455

443344

331133

232232

121121

},,{,:

},,{,:

},,{,:

},,{,:

},,{,:

WQQSO

WQQSO

WQQSO

WQQSO

WQQSO

Page 16: Query Aggregation for Providing Efficient Data Services in Sensor Networks

Weighted Zone-based Query Aggregation Algorithm (cont’d)

Q1(v1)

Q2(v2)

Q3(v3)

Q4(v4)

Q5(v5)

: Query region

Q: Input query

V: Attribute information

Consolidate overlapping zones in O

O1 O2

O3O4 O5

O1

3213211323121

554455

443344

},,,{,:

},,{,:

},,{,:

WWWQQQSSSO

WQQSO

WQQSO

Page 17: Query Aggregation for Providing Efficient Data Services in Sensor Networks

Weighted Zone-based Query Aggregation Algorithm (cont’d)

Q1(v1)

Q2(v2)

Q3(v3)

Q4(v4)

Q5(v5)

: Query region

Q: Input query

V: Attribute information

Sort the weights and assign queries to corresponding zone

O1 O2

O3O4 O5

O1

)( '

and allfor

first.ight largest weby OSort '

QQtrueBsQ

QQQQ

itt

ji

'

ti

55

454

413

12

11

451

,

,

}O,O,{O :Oset Sorted

OQ

OOQ

OOQ

OQ

OQ

Page 18: Query Aggregation for Providing Efficient Data Services in Sensor Networks

Weighted Zone-based Query Aggregation Algorithm (cont’d)

Q1(v1)

Q2(v2)

Q3(v3)

Q4(v4)

Q5(v5)

: Query region

Q: Input query

V: Attribute information

Calculate the access pointO1 O2

O3O4 O5

O1

Y)(X,point access central local its has and

S(size)by weight assigned isquery each

queries-subK

include O zonequery aggregatedAn

1 k,...,QQ

: access point

Q6

New aggregated queries

Query 1:{Q1, Q2, Q3}

Query 2:{Q4 ,Q5}

Page 19: Query Aggregation for Providing Efficient Data Services in Sensor Networks

Performance Evaluation-- Experimental Model

A grid-topology network 1500m x 1500m Grid size is 5m x 5m N queries, each of which is m-bit long Each query uniformly request the data from

area of S (=200). Query messages are combined with

compression ratio(0.7).

Page 20: Query Aggregation for Providing Efficient Data Services in Sensor Networks

Performance Evaluation-- Experimental Model (cont’d)

The energy consumption of sending message is calculated by

The energy consumption of receiving a message is calculated by

2***),( baEaEbaE fselecttx

aEbaE electrx *),(

2*100,50 mbitpJEbit

nJE fselect

Page 21: Query Aggregation for Providing Efficient Data Services in Sensor Networks

Performance Evaluation

Page 22: Query Aggregation for Providing Efficient Data Services in Sensor Networks

Performance Evaluation(cont’d)

Page 23: Query Aggregation for Providing Efficient Data Services in Sensor Networks

Performance Evaluation(cont’d)

Page 24: Query Aggregation for Providing Efficient Data Services in Sensor Networks

Conclusion

Query Aggregation A multi-layer overlay-based framework for

efficient sensor data service can support other routing protocols

An effective query aggregation mechanism do not consider the existing topology and distribution

of sensors query buffer