optimizing cascade data aggregation for vanets khaled ibrahim and michele c. weigle department of...

28
Optimizing CASCADE Dat a Aggregation for VANE Ts Khaled Ibrahim and Michele C. Weig le Department of Computer Science, Old Dominion Univer sity MASS 2008

Upload: tamsyn-day

Post on 18-Jan-2016

213 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Optimizing CASCADE Data Aggregation for VANETs

Khaled Ibrahim and Michele C. WeigleDepartment of Computer Science, Old Dominion University

MASS 2008

Page 2: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Outline

Introduction Related Work

CASCADE ( Cluster-based Accurate Syntactic Compression fo Aggregated Data in VANETs )

Motivation and Goal Proposed algorithm Analysis Conclusion

Page 3: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Introduction Vehicular Ad-hoc Networks (VANETs) have been proposed to provid

e drivers with advance notification of traffic congestion using wireless communication.

The more vehicles participating in the VANET the more messages are sent a frame size is finite

Recently, data aggregation in VANETs has much attention to reduce the frame size.

let a single frame to carry large number of information about vehicles

Page 4: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Related Work - CASCADE

CASCADE GLOBECOM Workshops, 2008 IEEE

This paper proposed a method for accurate aggregation of highway traffic information in VANETs. let a single frame to carry large number of information about vehicles.

CASCADE uses compression to let frame can carry more traffic information without losing accuracy.

Page 5: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Related Work - CASCADE

Assumptions Each vehicle is equipped with a GPS.

Each vehicle is also pre-assigned a public key’s certificate, used for authentication.

Cluster width is 16 m, cluster length is 64 m, the local view is 1600 m

Page 6: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Related Work - CASCADE

Each vehicle broadcasts a primary frame contains the its primary record every 300-400 ms.

The primary frame contains following:

Time-to-live ( TTL )

Page 7: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Related Work - CASCADE

A vehicle's local view is made up of primary records representing vehicles a certain distance ahead. ( local view = 1.6km )

A Vehicle ‘s Local view

a

Page 8: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Related Work - CASCADE

As primary frames are received To reduce the data size, the vehicle will grouped the records into their

corresponding clusters, based on their distance from the receiving vehicle.

a a

Page 9: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Related Work - CASCADE

To achieve compression Each vehicle is represented by the difference between it and the cluster

center.

XY

cluster

difference speedthebits S )5(

speedmedian the of s15m/s,15m/- the withinis

speedthe if indicates ) bits 2 Flag( Indicator Speed

Page 10: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Related Work - CASCADE

The total bits used to localize a vehicle is reduce to 16 bits.

XY

cluster4m

64m

difference speedthebits S )5(

speedmedian the of s15m/s,15m/- the withinis

speedthe if indicates ) bits 2 Flag( Indicator Speed

162514log12/64log 22

Page 11: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Related Work - CASCADE

The primary data for each vehicle (location and speed) is represented in 136 bits (17 bytes) while the compact data for each vehicle is represented in at most 16 bits.

162514log12/64log 22

Page 12: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Related Work - CASCADE

Once the Compact records for all vehicles in the local view have been created, an aggregated cluster record ( ACR ) is form for each cluster.

Page 13: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Related Work - CASCADE

Once the ACRs are constructed, they are concatenated into a aggregated frame and sent via broadcast.

100464/441600

100

Page 14: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Motivation The distance covered by the local view depends upon the number of vehicle

records that can fit in a single IEEE 802.11 frame (2312 bytes).

Local view

Page 15: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Motivation

If the value of Lc or Wc is risen, the compact record size for a vehicle will increase the data

The vehicle information in a frame will reduce, the local view is smaller

XY

clusterCW

CLLocal view

Page 16: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Goal

In IEEE 802.11 frame size constraint, we determine the optimal cluster size To maximize the local view length.

Page 17: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Proposed algorithm

ACR_1 ACR_2 … ACR_M

CR_1 CR_2 … CR_K

In determining the optimal cluster size, we strive to find an appropriate trade-off that will minimize the aggregated frame size and maximize the local view length.

ACR: Aggregated Cluster Record( the cluster data in local view )CR: Compact record( the vehicle data in a cluster )

Page 18: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Proposed algorithm

Aggregated frame Size Aggregated Cluster Record ( ACR ) Size

We assume that the width of one lane is 4 m and that the average vehicle length is 5 m.

LC

WC

Page 19: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Proposed algorithm

Aggregated frame Size Compact Record ( CR ) Size

XY

clusterm4

CL

difference speedthebits S )5( ) bits 2 Flag( Indicator Speed

112

4

mlog X.size 2

912

12

4

C

22

Llog

mlog CR.size

Page 20: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Proposed algorithm

The Aggregated frame size function

roadway the on lanes of number theC

4m is lane each of sizethe assum we

L

LC

Length

WC

Page 21: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Proposed algorithm

In std. frame size constraint

SectionData the withoutframe the of sizethe is sizeAF_nodata.bytes 2321 sizeMAC_Frame.

ata.size Max_ACR_d .sizeACR_Header ACR.size

LC

WC

ACR.count

Page 22: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Proposed algorithm

length clusterL

mW LC

C

CC

4/

LC

Length

WC

Page 23: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Analysis

In our analysis, we consider four different cluster lengths ( 62m, 126m, 254m, 510m ) and three different cluster widths ( 1 lane, 2lanes, 4lanes )

6 bits 7 bits 8 bits 9 bits

Page 24: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Analysis

For each traffic density, as the cluster dimensions change, the associated local view will change.

We consider 53 vehicles/km as low density, 66 vehicles/km as medium density, and 90 vehicles/km as high density.

N is total number of vehicles in the local viewM is the number of clusters in the local viewK is the maximum number of vehicles per cluster

Page 25: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Analysis Implies that increasing the cluster width to more than 4

lanes will provide no benefit

Page 26: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Analysis

We calculate the aggregated frame size for various cluster sizes and also considering different traffic densities.

Each vertical line represents the possible frame sizes for the specific cluster dimension.

Page 27: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Conclusion

In our analysis, we determined that a cluster size 16 m wide and 126 m long would provide the best trade-off between frame size and local view length.

Page 28: Optimizing CASCADE Data Aggregation for VANETs Khaled Ibrahim and Michele C. Weigle Department of Computer Science, Old Dominion University MASS 2008

Thank you