thesis presentation chayanin thaina advisor : asst.prof. dr. kultida rojviboonchai

83
Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Upload: marilynn-lynch

Post on 17-Jan-2016

215 views

Category:

Documents


1 download

TRANSCRIPT

Page 1: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Thesis Presentation

Chayanin Thaina

Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Page 2: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Outline

• VANETs

• Beaconing in VANETs

• Related work

• Proposed adaptive beaconing scheme

• Performance and Evaluation

• Conclusion

2

Page 3: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Outline

3

Page 4: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Vehicular Ad-Hoc Networks (VANETs)

• Intervehicle communication

• VANETs characteristics Nodes move with

high speed

Frequently change in network topology

High number of nodes

Vehicular Ad hoc Networks (VANETs)Avaliable from: http://www.car-to-car.org/

4

Page 5: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Outline

5

Page 6: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Beaconing in VANETs

• Vehicle Discover neighbors Exchange information

• Information may contain NodeID Position Direction Velocity Acknowledgement e.g.

6

Page 7: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Beaconing in VANETs

“Most of protocols in VANET using constant beaconing rate”

7

Page 8: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Examples of protocols (using constant beaconing rate)

• Routing protocol VADD Vehicle-assisted data delivery in vehicular Ad hoc networks (IEEE Trans. on vehicular tech., 2008)

• Broadcasting protocol AckPBSM Acknowledge Parameterless broadcast Protocol in static to highly mobile ad hoc networks (VTC, 2009) DV-Cast Distributed Vehicular Broadcast Protocol for Vehicular Ad-hoc Networks(IEEE Wireless communication, 2010)

Beacon interval

0.5 s

0.5 s

1 s

8

Page 9: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Outline

9

Page 10: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Related work

CAR : Connectivity-Aware Routing in Vehicular Ad Hoc Networks

(Valery Naumov and Thomas R. Gross, Infocom 2007)

Improving Neighbor Localization in Vehicular Ad Hoc Networks to Avoid Overhead from Periodic Messages

(Azzedine Boukerche, Cristiano Rezende and Richard W. Pazzi ,GLOBECOM 2009)

Efficient Beacon Solution for Wireless Ad-Hoc Networks (Nawut Na Nakorn and Kultida Rojviboonchai, JCSSE 2010)

Exploration of adaptive beaconing for efficient intervehicle safety communication (Robert K. Schmidt, Tim Leinmuller, Elmar Schoch, Frank Kargl and Gunter Schafer, IEEE Network, 2010)

Page 11: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Connectivity-Aware Routing in Vehicular Ad Hoc Networks (CAR)

• Methodology Beaconing interval is changed according to the

number of neighbors

Calculate beacon interval

0.5Beacon Interval weight

11

weight : A weight proportional to the number of neighbors

Page 12: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Improving Neighbor Localization in Vehicular Ad Hoc Networks to Avoid Overhead from Periodic Messages

• Methodology Beacon rate adaptation based on differences in predicted

position

Use last beacon message to estimate position

Send next beacon- When the difference between the predicted and actual position is greater than threshold value

12

Page 13: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Efficient Beacon Solution for Wireless Ad-Hoc Networks

• Methodology Adapt beacon based on number of neighbors and

number of buffered messages

1 2( ) ( )s w n w m

s : Dense value, n : Number of neighbors, m : Number of buffer messages

w1, w2 : Weight value of number of neighbors and number of buffer messages

13

Page 14: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Efficient Beacon Solution for Wireless Ad-Hoc Networks

LIA : Linear Adaptive Algorithm

STA : Step Adaptive Algorithm

(3)

14

Page 15: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Exploration of adaptive beaconing for efficient intervehicle safety communication

• Methodology Adjust the beacon frequency dynamically to the current

traffic situation

15

Page 16: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

The drawbacks of previous work

• Some works have to use so many tests to find the constant value for adjusting beacon interval.

• Some works, vehicles need GPS data for adjusting beacon interval.

16

Page 17: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Conclusion of related workCAR Improving

Neighbor Localization in VANETs to Avoid Overhead from Periodic Messages

Efficient Beacon Solution for Wireless Ad-Hoc Networks

Exploration of adaptive beaconing for efficient intervehicle safety communica-tion

Proposed(Linear regression analysis)

Proposed(k-Nearest Neighbor)

Proposed(LIA+NCR)

Parameters used in calculation

- Number of neighbors

- Position- Speed- Direction

- Number of neighbors- Number of messages

- Velocity- Acceleration- Yaw rate- Emergency/ Regular vehicle- Vehicle density- Special situation

- Number of neighbors- Number of messages- Speed of Data dissemina- tion

- Number of neighbors- Number of messages- Speed of Data dissemina- tion

- Number of neighbors- Number of messages- Neighbor changing rate

Selection mechanisms

Linear function

Predicted position

- Linear Adaptive Algorithm (LIA)- Step Adaptive Algorithm (STA)

X

- Linear regression analysis

- Instance- Based Learning

Linear function

Page 18: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Conclusion of related workCAR Improving

Neighbor Localization in VANETs to Avoid Overhead from Periodic Messages

Efficient Beacon Solution for Wireless Ad-Hoc Networks

Exploration of adaptive beaconing for efficient intervehicle safety communica-tion

Proposed(Linear regression analysis)

Proposed(k-Nearest Neighbor)

Proposed(LIA+NCR)

GPS X X X X X

Beacon interval

>=0.5 X 1.5-7 X >=2.1509 1.5-9 >=1.5

Page 19: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Outline

19

Page 20: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Goals of our adaptive beaconing schemes

• Reduce beacon overhead

• Maintain Reliability Retransmission overhead

• Provide the speed of data dissemination according to the requirement of each application

20

Page 21: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Design of our adaptive beaconing schemes

• A study on adaptive beaconing is divided into 3 parts

1. Study on the parameters which affect adaptive beacon rate1. Study on the parameters which affect adaptive beacon rate

3. Study on the methods that can be applied on adaptive beacon rate 3. Study on the methods that can be applied on adaptive beacon rate

21

2. Study on the system performance when using constant beacon rate and different parameters2. Study on the system performance when using constant beacon rate and different parameters

Page 22: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Node’s environment- Number of neighbors

- Number of buffered messages

Application requirement- Speed of data dissemination

Design of our adaptive beaconing schemes

1. Study on the parameters which affect adaptive beacon rate1. Study on the parameters which affect adaptive beacon rate

Number of neighbors +Number of messages

High

Beacon rate

Low

Number of neighbors +Number of messages

Low

Beacon rate

High

22

• A study on adaptive beaconing is divided into 3 parts

Page 23: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Test sending beacon with different beacon intervals and different node’s environment.

Gather all the results and conclude the appropriate beacon intervals.

Design of our adaptive beaconing schemes

2. Study on the system performance when using constant beacon rate and different parameters

2. Study on the system performance when using constant beacon rate and different parameters

23

• A study on adaptive beaconing is divided into 3 parts

Metrics

-Beacon overhead

-Reliability

-Retransmission overhead

-Speed of data dissemination

Page 24: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

2. Study on the system performance when using constant beacon rate and different parameters

24

Beacon overhead

Highway Scenarios Urban Scenarios

Beacon rate --> Beacon overhead

Page 25: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

2. Study on the system performance when using constant beacon rate and different parameters

25

Reliability

Highway Scenarios Urban Scenarios

Beacon rate in Dense area --> Reliability

Beacon rate in Sparse area --> Reliability

Page 26: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

2. Study on the system performance when using constant beacon rate and different parameters

26

Retransmission overhead

Highway Scenarios Urban Scenarios

Beacon rate --> Retransmission

Page 27: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

2. Study on the system performance when using constant beacon rate and different parameters

27

Speed of data dissemination (Low density 2 veh/km)

HighwayScenarios

UrbanScenarios

Sparse area --> Beacon rate

Page 28: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

2. Study on the system performance when using constant beacon rate and different parameters

28

Speed of data dissemination (Medium density 30 veh/km)

HighwayScenarios

UrbanScenarios

Page 29: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

2. Study on the system performance when using constant beacon rate and different parameters

29

Speed of data dissemination (High density 80 veh/km)

HighwayScenarios

UrbanScenarios

Dense area --> Beacon rate

Page 30: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

2. Study on the system performance when using constant beacon rate and different parameters

30

Gather all the results and conclude the appropriate beacon intervals

- Type of scenario that is suitable for choosing is the highway scenario

- In this study, considering the speed of data dissemination in highway to be within 10 s.

Page 31: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

2. Study on the system performance when using constant beacon rate and different parameters

31

Appropriate beacon intervals

Density (veh/km) Beacon interval (s.)

2 1.56 3

10 720 730 940 960 980 9

Page 32: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Method that determines a statistical model

Machine Learning technique

Improve the solution of Linear Adaptive Interval (LIA)

Design of our adaptive beaconing schemes

3. Study on the methods that can be applied on adaptive beacon rate 3. Study on the methods that can be applied on adaptive beacon rate

Linear regression analysis

k-Nearest Neighbor (k - NN)

k-Nearest Neighbor (k - NN)

32

• A study on adaptive beaconing is divided into 3 parts

LIA with Neighbor Change Rate (LIA+NCR)

LIA with Neighbor Change Rate (LIA+NCR)

Page 33: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Linear regression analysis

• Finding relationship between independent variables and a dependent variable

Y a bX

: Dependent variable (Beacon Interval)

: Independent variable (Number of neighbors + number of messages)

: Regression coefficients

Y

X

,a b

1

2

1

( )( )

( )

n

i ii

n

ii

x x y yb

x x

,a y bx

: average of all recorded , : average of all recorded x x y y

33

Page 34: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

k-Nearest Neighbor

• Instance-based learning

• Training examples will be collected in the form of

• Assume all instances corresponding to points in the n-dimensional space

• Define k value which denotes the number of nearest neighbors

34

))(,( ii xfx

Page 35: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

k-Nearest Neighbor

• If has query instance - Nearest neighbors are defined by Euclidean distance

35

qx

: distance between and

: the value of the th attribute of instance

qx

( )r ia x

Page 36: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Weigh each k-nearest neighbor according to their distance to the query point qx

: distance between and qx

: weight value of each k instance

36

k-Nearest Neighbor

Page 37: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Output

: weight value of each k instance

37

k-Nearest Neighbor

Page 38: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Improve the solution of Linear Adaptive Interval (LIA)

• Using a new parameter, “neighbor change rate” to improve the previous adaptive solution call “Linear Adaptive Algorithm” (LIA)

38

Neighbor nodes --> Beacon rateNeighbor nodes --> Beacon rate

Page 39: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Improve the solution of Linear Adaptive Interval (LIA)

• Improve the solution of Linear Adaptive Algorithm (LIA) by using neighbor change rate (NCR) divided into 3 parts

Neighbor Change Rate (NCR) Using only the data of neighbor change rate to adapt beacon interval

Linear Adaptive Algorithm with Neighbor Change Rate (limited) (LIA+NCR(limited))Using the data of neighbor change rate and network density to adapt beacon interval (Limited the maximum beacon interval)

Linear Adaptive Algorithm with Neighbor Change Rate (unlimited) (LIA+NCR(unlimited))Using the data of neighbor change rate and network density to adapt beacon interval (unlimited the maximum beacon interval)

39

Page 40: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Improve the solution of Linear Adaptive Interval (LIA)

Neighbor Change Rate (NCR)

(LIA+NCR (limited)) (LIA+NCR (unlimited))

Neighbor changing rate (NCR)

Neighbor changing rate (NCR)

Neighbor changing rate (NCR)

( ) min( ( ), )f s MinInv c NCR MaxInv ( ) min( ( ( )), )f s MinInv c s NCR MaxInv ( ) ( ( ))f s MinInv c s NCR

Neighbor changing rate (NCR)- When the number of neighbor nodes increase NCR + 1- When the number of neighbor nodes decrease NCR – 1

1 2( ) ( )s w n w m

n : Number of neighbors,

m : Number of buffer messages

w1, w2 : Weight value

n : Number of neighbors,

m : Number of buffer messages

w1, w2 : Weight value

1 2( ) ( )s w n w m

Network density Network density

Page 41: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Example

41

Training data for adaptive algorithms

Network density Beacon interval (s.)

1 1.53 35 7

10 715 920 930 940 9

Page 42: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

42

Example (Linear regression analysis)

Page 43: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Example (Linear regression analysis)

43

Ex. - If node has 3 neighbor nodes and 1 buffered messages

- Dense value = 3+1 = 4

- Next beacon interval

ˆ 2.1509 0.4957

ˆ 2.1509 (0.4957 4)

Y X

Y

= 4.1337

Page 44: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Each node will contain a table that collects the training examples ))(,( ii xfx

44

Example (k-Nearest Neighbor)

Network density(xi)

Beacon intervalf(xi)

x1 1 1.5

x2 3 3

x3 5 7

x4 10 7

x5 15 9

x6 20 9

x7 30 9

x8 40 9

Page 45: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Define k value (denotes the number of the nearest neighbors)

45

Example (k-Nearest Neighbor)

k = 2 Network density

(xi) Beacon interval

f(xi)

x1 1 1.5

x2 3 3

x3 5 7

x4 10 7

x5 15 9

x6 20 9

x7 30 9

x8 40 9

Page 46: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Network density(xi)

Beacon intervalf(xi)

x1 1 1.5

x2 3 3

x3 5 7

x4 10 7

x5 15 9

x6 20 9

x7 30 9

x8 40 9

Ex. - If node has 3 neighbor nodes and 1 buffered messages- Dense value (xq) = 3+1 = 4

46

Example (k-Nearest Neighbor)

Calculate the distance between xq and xi

2

1( , ) (4 1) 3qd x x 2

2( , ) (4 3) 1qd x x 2

3( , ) (4 5) 1qd x x 2

4( , ) (4 10) 6qd x x

2

5( , ) (4 15) 11qd x x

2

6( , ) (4 20) 16qd x x

2

7( , ) (4 30) 26qd x x 2

8( , ) (4 40) 36qd x x

Page 47: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Network density(xi)

Beacon intervalf(xi)

x1 1 1.5

x2 3 3

x3 5 7

x4 10 7

x5 15 9

x6 20 9

x7 30 9

x8 40 9

Ex. - If node has 3 neighbor nodes and 1 buffered messages- Dense value (xq) = 3+1 = 4

47

Example (k-Nearest Neighbor)

Calculate the distance between xq and xi

2

1( , ) (4 1) 3qd x x 2

2( , ) (4 3) 1qd x x 2

3( , ) (4 5) 1qd x x 2

4( , ) (4 10) 6qd x x

2

5( , ) (4 15) 11qd x x

2

6( , ) (4 20) 16qd x x

2

7( , ) (4 30) 26qd x x 2

8( , ) (4 40) 36qd x x

Page 48: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Network density(xi)

Beacon intervalf(xi)

x1 1 1.5

x2 3 3

x3 5 7

x4 10 7

x5 15 9

x6 20 9

x7 30 9

x8 40 948

Example (k-Nearest Neighbor)

2

2

2 2 2

2

( , ) (4 3) 1

1 11

( , ) 1

q

q

d x x

wd x x

2

3

3 2 2

3

( , ) (4 5) 1

1 11

( , ) 1

q

q

d x x

wd x x

Calculate the weight value of each nearest neighbor

Page 49: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Network density(xi)

Beacon intervalf(xi)

x1 1 1.5

x2 3 3

x3 5 7

x4 10 7

x5 15 9

x6 20 9

x7 30 9

x8 40 949

Example (k-Nearest Neighbor)

2 2 2

2

1 11

( , ) 1q

wd x x

3 2 2

3

1 11

( , ) 1q

wd x x

Calculate the output

(1 3) (1 7)ˆ ( )1 1qf x

= 5

Page 50: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

50

Example (Neighbor change rate (NCR))

Ex. Previous- Node has 3 neighbor nodes and Neighbor Change Rate (NCR) is 2

Current- 2 Neighbor nodes adding- Node calculates the neighbor change rate (NCR) NCR+1 = 3

Calculate the next beacon interval

( ) min( ( ), )f s MinInv c NCR MaxInv MinInv = 1.5, MaxInv : 7,

c = 0.2

( ) min(1.5 (0.2 3),7)f s = 2.1

Page 51: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

51

Example (LIA+NCR (limited))

Ex. Previous- Node has 25 neighbor nodes, 5 buffered messages and Neighbor Change Rate (NCR) is 5

Current- 3 neighbor nodes leaving - Node calculates the neighbor change rate (NCR) NCR-1 = 4- Calculate the network density = 22+5 = 27

Calculate the next beacon interval

( ) min( ( ( ), ))f s MinInv c s NCR MaxInv MinInv = 1.5, MaxInv : 7,

c = 0.2

( ) min(1.5 0.2 (27 4),7)f s = 7

Page 52: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

52

Example (LIA+NCR (unlimited))

Ex. Previous- Node has 25 neighbor nodes, 5 buffered messages and Neighbor Change Rate (NCR) is 5

Current- 3 neighbor nodes leaving - Node calculates the neighbor change rate (NCR) NCR-1 = 4- Calculate the network density = 22+5 = 27

Calculate the next beacon interval

( ) ( ( ))f s MinInv c s NCR MinInv = 1.5, MaxInv : 7,

c = 0.2

( ) 1.5 0.2 (27 4),7)f s = 7.7

Page 53: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Outline

53

Page 54: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Performance and Evaluation

• Case study

DECA : Density-Aware Reliable Broadcasting in

Vehicular Ad Hoc Networks

(ECTI-CON, 2010)

54

Page 55: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

DECA : Density-Aware Reliable Broadcasting in Vehicular Ad Hoc Networks

• Reliable broadcast protocol

• Store and forward solution

• Exchange beacon message Beacon information contains

Use Linear Adaptive Algorithm : LIA

55

Node Identifier(4 bytes)

Number of neighbors(1 byte)

Message Ack#1 #2…

Page 56: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

DECA : Density-Aware Reliable Broadcasting in Vehicular Ad Hoc Networks

• Broadcast message Sender select the forwarder from its neighbor list - Neighbor with the highest density will be selected

Selected node rebroadcast the message immediately

Other neighbors (which are not selected) - Store the message and set waiting timeout

In case the selected node doesn’t rebroadcast the message- Other neighbors will rebroadcast the message

56

Page 57: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Simulation Setup

• Network Simulation : NS-2.34

• Traffic Simulation Trace generator : SUMO (Simulation of Urban

MObility)

XML convertor to NS2 trace : TraNS

57

Page 58: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

• Scenario 3x3 km. with 2 lanes

Urban Scenario

4 km. with 4 lanes

Highway Scenario

58

Simulation Setup

Page 59: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Simulation Setup

Broadcasting message 1,5,10,15

Vehicle density Highway : 6,10,20,30,40,60,80 veh/kmUrban : 2,10,30,60,80 veh/km

Maximum speed Highway : 50,80 km/hUrban : 120 km/h

Packet life time Highway : 10 s.Urban : 50 s.

Linear Adaptive Algorithm (LIA)

Beacon interval : 1.5-7 ; (c = 0.2, MinInv = 1.5, MaxInv = 7)

Linear regression analysis Regression coefficients : a = 2.1509, b = 0.4957

k-Nearest Neighbor (k-NN) Number of nearest neighbor (k) = 2

LIA+NCR (limited) Beacon interval : 1.5-7; (c = 0.2, MinInv = 1.5, MaxInv = 7)

LIA+NCR (unlimited) Beacon interval : >=1.5; (c = 0.2, MinInv = 1.5)

Requirement of speed of data dissemination

Highway : 10 s.Urban : 15 s.

59

Page 60: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

• Use DECA to evaluate 6 beaconing schemes

LIA : Linear Adaptive Algorithm

Linear regression : Linear regression analysis

k-NN : k-Nearest Neighbor NCR : Neighbor Change Rate LIA+NCR (limited) : Linear Adaptive Algorithm with

Neighbor Change Rate (limited maximum beacon interval)

LIA+NCR (unlimited) : Linear Adaptive Algorithm with Neighbor Change Rate (unlimited maximum beacon interval)

60

Simulation

Page 61: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Simulation

• Metrics Beacon overhead

- bandwidth that has been used for every beacon (bytes/node/message)

Reliability- percentage number of received node to number of total node

Retransmission overhead- bandwidth that has been used for data transmission (bytes/node/message)

Speed of data dissemination - percentage of number of node that received message at time (t)

61

Page 62: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Simulation

• Metrics Number of beacon

- The number of beacon that has been sent in scenario Number of retransmission

- The number of data transmission that has been broadcast in scenario

62

Page 63: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

• LIA, Linear Regression, k-Nearest Neighbor - Broadcasting message : 10 messages

Simulation result (Beacon Overhead)

Highway Scenarios Urban Scenarios

k-Nearest Neighbor can reduce beacon overhead up to 54% in highway and 41% in urban scenarioLinear regression can reduce beacon overhead up to 78% in highway and 70% in urban scenario

Page 64: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

• LIA, NCR, LIA+NCR (limited), LIA+NCR (unlimited) - Broadcasting message : 10 messages

Simulation result (Beacon Overhead)

Highway Scenarios Urban Scenarios

LIA+NCR (limited) can reduce beacon overhead up to 18% in highway and 11% in urban scenarioLIA+NCR (unlimited) can reduce beacon overhead up to 50% in highway and 51% in urban scenario

Page 65: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

• LIA, Linear Regression, k-Nearest Neighbor - Broadcasting message : 10 messages

Simulation result (Reliability)

67

Highway Scenarios Urban Scenarios

Page 66: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

• LIA, NCR, LIA+NCR (limited), LIA+NCR (unlimited) - Broadcasting message : 10 messages

Simulation result (Reliability)

68

Highway Scenarios Urban Scenarios

Page 67: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

• LIA, Linear Regression, k-Nearest Neighbor - Broadcasting message : 10 messages

Simulation result (Retransmission Overhead)

69

Highway Scenarios Urban Scenarios

Page 68: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

• LIA, NCR, LIA+NCR (limited), LIA+NCR (unlimited) - Broadcasting message : 10 messages

70

Highway Scenarios Urban Scenarios

Simulation result (Retransmission Overhead)

Page 69: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

• LIA, Linear Regression, k-Nearest Neighbor - Broadcasting message : 10 messages

Simulation result (Speed of data dissemination)

Highway Scenarios

Urban Scenarios

Low density : 10 veh/km

Page 70: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

• LIA, Linear Regression, k-Nearest Neighbor - Broadcasting message : 10 messages

Simulation result (Speed of data dissemination)

Highway Scenarios

Urban Scenarios

Medium density : 30 veh/km

Page 71: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

• LIA, Linear Regression, k-Nearest Neighbor - Broadcasting message : 10 messages

Simulation result (Speed of data dissemination)

Highway Scenarios

Urban Scenarios

High density : 80 veh/km

Page 72: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

• LIA, NCR, LIA+NCR (limited), LIA+NCR (unlimited) - Broadcasting message : 10 messages

Simulation result (Speed of data dissemination)

Highway Scenarios

Urban Scenarios

Low density : 10 veh/km

Page 73: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

• LIA, NCR, LIA+NCR (limited), LIA+NCR (unlimited) - Broadcasting message : 10 messages

Simulation result (Speed of data dissemination)

Highway Scenarios

Urban Scenarios

Medium density : 30 veh/km

Page 74: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

• LIA, NCR, LIA+NCR (limited), LIA+NCR (unlimited) - Broadcasting message : 10 messages

Simulation result (Speed of data dissemination)

Highway Scenarios

Urban Scenarios

High density : 80 veh/km

Page 75: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

• LIA, Linear Regression, k-Nearest Neighbor - Broadcasting message : 10 messages

Simulation result (No.Beacon&No.Retransmission)

77Highway Scenarios

Page 76: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

• LIA, Linear Regression, k-Nearest Neighbor - Broadcasting message : 10 messages

Simulation result (No.Beacon&No.Retransmission)

78Urban Scenarios

Page 77: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

• LIA, LIA+NCR (limited), LIA+NCR (unlimited) - Broadcasting message : 10 messages

Simulation result (No.Beacon&No.Retransmission)

79Highway Scenarios

Page 78: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

• LIA, LIA+NCR (limited), LIA+NCR (unlimited) - Broadcasting message : 10 messages

Simulation result (No.Beacon&No.Retransmission)

80Urban Scenarios

Page 79: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Outline

81

Page 80: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Conclusion

• Propose 3 adaptive beaconing methods Linear regression analysis k-Nearest Neighbor Improve the solution of Linear Adaptive Algorithm (LIA)

by using neighbor change rate (NCR)

• 2 methods can be applied to adjust beacon interval according to Node’s environment Application requirement

82

Lowest beacon overhead

Page 81: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Conclusion

• Our proposed methods can save bandwidth Highway : 78 % Urban : 70%

• Our proposed methods can maintain Reliability Speed of data dissemination

83

Page 82: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

Question &

Answer

84

Page 83: Thesis Presentation Chayanin Thaina Advisor : Asst.Prof. Dr. Kultida Rojviboonchai

THANK YOU.

85