Vehicular Density-Dependent Data

8/8/2019 Vehicular Density-Dependent Data

http://slidepdf.com/reader/full/vehicular-density-dependent-data 1/6

VD4: Vehicular Density-dependent Data

Delivery Model in Vehicular Ad hoc Networks

Avijit Gupta1, Vineet Chaudhary2, Vivek Kumar 3, Bharat Nishad4, Shashikala Tapaswi5

ABV-Indian Institute of Information Technology & Management Gwalior, [email protected], 2

[email protected], [email protected], [email protected], [email protected]

Abstract — In this paper, we present the VD4 model for effective

delivery of data packets in vehicular ad hoc networks. The

major challenges in such networks are disconnections, high

mobility and resource poor environment. For providing data

delivery under such constraints we adopt the method of carry

and forward in which a vehicle transfers its data to another

vehicle if it is moving towards the destination. For optimizingperformance in such situations, we can make a probabilistic

estimation as to which path the packet should follow when there

are multiple options available. One such estimation can be

based on road density and the average packet delay on the

selected road. The paper focuses on the development of such a

model that can act as a solution to reduce the packet delivery

delay and improve the data delivery ratio.

Keywords-Vehicular Ad hoc Networks; Road Side Unit; routing

protocol; carry and forward; vehicular density

I. I NTRODUCTION

A Vehicular Ad hoc Network (VANET) is a

communication technology that creates network between

vehicles [1] or roadside gateways to allow information

exchange between users. The importance of vehicular ad

hoc networks is increasing day by day and it is estimated

that they will play a key role in providing transportation in

the near future. Its major application lies in providing safety,

comfort, and critical information to drivers on road. Due to

the increasing concern in this field, many research works are

targeted towards more and more optimization in this area.

Vehicular Ad hoc Networks are useful in scenarios where

a driver wants information, even when he is miles from thedestination. As an example, if he wants to view the traffic

pattern of roads to his destination so as to choose the path

with minimum congestion, he can request for the same even

when he is far from the destination. In such applications, the

user can tolerate some delay. Such a service is difficult to be

provided if the cost is high or the infrastructure is damaged.

VANETs differ considerably from Mobile Ad hoc Networks

(MANETs) because the nodes in VANETs are much faster,

have different mobility patterns and are generally confined

to road maps, which make location estimation much easier

as compared to MANETs. Therefore, the protocols suitable

for MANETs may not necessarily be suited for VANETs

and can be optimized to provide better results. The network

connection in VANETs is generally achieved by Road SideUnits (RSUs) that can either act as router for vehicles to

access the network or just as buffer points(or data island)

between vehicles [11]. While the first option is costly, the

second option is prone to excessive delay as the destination

vehicle might never pass the RSU. In such a situation,

communication between RSUs can be accomplished to

provide connectivity to the destination vehicle, which is

again, expensive. In a resource poor environment, the

vehicles cannot always rely upon RSUs especially if they

are placed very far away. For removing the reliability of

vehicular data transfer on RSUs, we can use the carry and

forward [12] method in which a packet between RSUs can be carried by vehicles themselves. A similar approach was

followed in [13], where data is poured by a data center

along the nodes and they are delivered not only to the

vehicles on these nodes but also to the vehicles on the

intersecting nodes when they move across the intersection.

This type of broadcast is expensive and data is transmitted

to nodes which might not be on the path of the destination at

all.

II. STATE OF THE ART

Many researches have been done on vehicle

communication. Some papers have discussed on mobilitymodeling and optimization of (MAC) issues have been

deliberated upon [4][5]. Many researchers have focused

upon development and routing protocols [6][7].

Transportation safety has been discussed in [8][9] where

inter vehicle communication is achieved with static network

nodes. For comfort and entertainment, real time video

streaming between vehicles has been studied in reference

[10].

2010 Sixth Advanced International Conference on Telecommunications

978-0-7695-4021-4/10 $26.00 © 2010 IEEE

DOI 10.1109/AICT.2010.80

285

2010 Sixth Advanced International Conference on Telecommunications

978-0-7695-4021-4/10 $26.00 © 2010 IEEE

DOI 10.1109/AICT.2010.80

286



Multi hop data delivery in VANETs is challenging

primarily because of high mobility and sparse situations that

might be prevalent in areas where the service is required. As

the number of nodes on any road might vary from time to

time, it is extremely difficult to find end to end connectivity

for sparsely connected network. In such situations the

moving vehicles might be used as data carriers.

Furthermore, the data packets may be transferred from

vehicle to vehicle whenever a vehicle moving towards the

destination reaches in the vicinity of a vehicle which was

originally carrying the packet. Thus, through relays and

carry and forward, the packet can be transferred even when

no end to end connectivity is present between the source and

destination.

This paper deals with effective data delivery in

VANETs. We suggest a model to effectively route the

request for the service to the destination and receive the

reply within a tolerable delay time. Any packet might becarried by vehicles to the destination via multiple paths. But

for effective and efficient data delivery, we need to consider

the path in which the data transmission delay is tolerable.

Therefore, at every intersection, the data packet can be

transferred to the RSU present there which can be

transferred to a vehicle which moves along the optimal road

chosen according to maximum vehicular densities and

minimum delay.

The rest of this paper is organized as follows. Section III

describes the assumptions that need to be made to represent

the VD4 model. Section IV describes how to model data

delivery delay based on the VD4 model. Section Vdescribes the VD4 algorithm for vehicular ad hoc networks.

The evaluation of the model is done in Section VI; Section

VII concludes the paper.

III. ASSUMPTIONS

This section states the various assumptions that need to

be made to adequately represent the VD4 model. We

assume that for being a part of the VANET, the vehicles are

sufficiently equipped with wireless transmitters which can

transmit in a short range (100 m-200 m) for transmitting

data packet whenever a vehicle reaches the vicinity of theoriginal data packet carrying vehicle or to the RSU. Every

roadside intersection is equipped with a RSU, which is

capable of storing data packets sent by vehicles as and when

required. The information necessary for the proper routing is

included by the source in the packet at the time of

transmission. Each vehicle and RSU knows its present

location using GPS. This is already available in advanced

vehicles. Vehicles and RSUs also contain the information

about paths that might be statically stored as scaled maps.

The RSU maintains the information of vehicles such as the

speed, direction etc. that passed it and also an estimate of

total number of vehicles present on each path at a given

point in time. This information is periodically updated so

that the evaluation of the optimal path may be done with the

latest information. The vehicles are assumed to move with

uniform speed on a path.

IV. THE VD4 MODEL

The Vehicular Density-dependent Data Delivery Model is

based upon routing data packets based on the calculation of

the successor path which is the path of minimum delay from

the set of paths present on the intersection. For any path to

achieve minimum delay, it must be such that the vehicles

can (i) maximize wireless transmissions so that the time

spent by the packet carried by the vehicle is minimized (ii)

the average speed of vehicle movement on the path (the

average of speed of all the vehicles on the path) must be

high so as to reduce the packet delay.

At any time, any path at an intersection can be divided

into two parts. In the first part, the transmission of data

packets can be done wirelessly and hence quickly and the

other in which the vehicle has to carry the data packet itself.

Since it is difficult to predict the configuration of VANET

after a period of time, therefore that path which gives the

largest distance that can accomplished via wireless

transmission from the source RSU will be the best present

path for transmission because it will be a path of minimumdelay. Consider two RSUs at i and j. Suppose k is a point in

between i and j from where the vehicle has to carry the

packet manually as there are no vehicles in front of it to

transfer the packet to. The basic equation for calculating the

delay for each path at an intersection between two RSUs, i

and j can be written as:

(1)

where:α = Fraction of the path length where data packet can be

transmitted via wireless transmission ( )

β = Wireless Transmission delay per unit distance

lij = The length of the path between the RSUs at i and j

vij = The average speed vehicles on the path from RSUs at i

and j

286287



δ = Constant used for adjusting delay in case wireless

transmission takes place.

ρij = Vehicular density at the path between the RSUs at i

and j

The above equation is similar to the delay equation

obtained for the VADD Model [6], difference being that in

the aforementioned model, the data transmitted between two

intersection points is either completely wirelessly or

completely carried by vehicles. This might not be true when

intersection points are far from each other or the vehicles

are distributed irregularly in the path. Therefore the path is

divided into two parts, the first being the part where data

can be transmitted wirelessly and quickly. The fraction of

such path is represented by α. In the remaining fraction (1-

α) of the path, the vehicle must carry the packet.

Representing delay in this manner leads to more accuracy.

This can be effectively depicted by Figure 1.

Figure 1 – Depiction of how packet is carried partially by avehicle and partially wirelessly transmitted

After the RSU at the intersection knows the delays of

every individual path, it has to choose which path it should

use to forward the packet. The application of Dijkstra’s

Algorithm is not feasible since information at each RSU is

dynamic. The most optimal path at a particular time might

not remain the same for a long duration of time. Therefore,

computation of the complete path is not feasible.

As a solution to this problem we use the stochastic model

of VADD Model [6]. The expected delay according to the

VADD model for the packet from intersection point Im to In

can be written as:

(2)

where,

Di: The expected packet delivery delay from Ii to the

destination if the packet carrier at Ii uses road rij to

deliver the packet.

Pij: The probability that the packet is forwarded through

road rij at Ii.

N(j): The set of neighboring intersections of Ij

V. THE VD4 ALGORITHM

In this section, we introduce the VD4 algorithm for

optimal routing of the data packets to the destination.

Each time a vehicle passes a RSU, the following

information is provided to it: (i) The time of arrival of the

vehicle (as a timestamp) (ii) The speed of the vehicle (iii)

The direction of movement (iv) Data packets (in case it

carries them). The data packets that are received carry a

unique sequence number which is checked for duplicity at

the RSU. If the packet is already present at the RSU, it is

dropped otherwise it is forwarded to the farthest in range

vehicle on the most optimal path as has been calculated by

the delay model. Consider a situation in which a packet is

carried by a vehicle from RSUi through the most optimal

path to RSUj. The transmission of packet by RSUi

continues till an acknowledgement is received or the

timeout is reached. Once the forwarded packet reaches

RSUj at the other intersection point, the acknowledgement

packet is sent using VANET itself to RSUi, which stops

transmitting packets from then on and deletes it from its

memory. Duplicated packets are ignored by RSUj using the

sequence number of the received packet. Following this

approach, the packet is transmitted till the destination nodeis reached. The basic algorithm can be summarized as

follows:

Notations:

Ii : The Intersection under consideration.

Ri : The transmission range of roadside unit(RSU) at Ii.

Vi : The vehicle under consideration.

Pi : The packet carried by Vi.

Di : The direction towards which vehicle Vi is heading.

Dj : The destination direction of packet Pi.

Vj : The vehicle farthest in direction Dj such that its

distance from RSU at Ii <= Ri.

N[] : The outgoing roads at Ii.

Description:

1. For each vehicle Vi that reaches RSUi, do :

2. if vehicle entry already exists then

update speed, direction and timestamp for Vi.

else

287288



make new entry for speed, direction and time of

arrival for Vi.

end if

3. If Vi is carrying any packet, Pi then

If Pi is a packet with ACK mark, then

delete Pi from memory of RSU at Ii.

else if Di is equal to Dj then

Transmit Pi from RSUi back to Vi.

Mark Pi as SENT.

else

Vj = SELECT(Dj , Ri).

►Select Vehicle to carry the packet

Give Pi to Vj

Mark Pi as SENT.

end if

end if

4. Leave the intersection Ii.

►Procedure to select the vehicle which will further carry

the data packet.

Procedure SELECT( Di , Ri)

1. j := 0, D = INF, p := 0

2. while j <= n(N) do

dij =

calculate

if D >= Dij then

D = Dij p = j

end if

3. Select farthest vehicle on Road N(p) whose

distance from Ii <= Ri.

4. return the selected vehicle.

The basic approach of the algorithm is to develop a data

delivery model such that packets can be delivered from

source to the destination using the intermediate RSUs. Once

a car reaches a RSU, the packets it carries is checked. If the

packet is an acknowledgement packet, then the packet is

deleted from the memory of RSU. In case it is not present in

the memory of RSU, it is incorporated in the memory and

transmitted to the farthest vehicle travelling towards the

destination. The optimal successor path is selected using the

delay equations stated above. The packets are delivered to

all vehicles moving towards the destination on the same

path until one of the packets is delivered to the RSU at next

hop. When this happens, an acknowledgement packet is sent

from that RSU to the source RSU via vehicle travelling in

that direction. Once the acknowledgement packet reaches

the source RSU, the packet transmission stops and packet is

deleted from the memory of RSU. This continues till the

packet is received by the destination.

VI. PERFORMANCE EVALUATION

In this section, we evaluate the performance of the VD4

protocol. We compare the performance of VD4 with several

protocols like Dynamic Source Routing (DSR)[15], the

epidemic routing protocol [16] , Greedy Perimeter Stateless

Routing (GPSR) [16] and Hybrid – Vehicle Assisted Data

Delivery (H-VADD) [6]. We use the simulationenvironment provided by NCTUns 6.0, which is widely

used to simulate VANETs. The following are the

parameters used during the simulation:

Simulation Area 10000m X 8000m

No of vehicles 180

Number of Intersections 5-50

CBR rate .1 – 1 packet per second

Vehicle Velocity 40km/hr - 60 km/hr

Communication Range 200m

Data packet size 1 KB

A snapshot of a limited section of the simulation area is

shown in Figure 2.

ITS Cars are randomly placed on the paths to depict an

environment. The vehicles move in simulation

environments with a speed of 40km/hr – 60km/hr, which

may vary according to restrictions on a particular path.

Various simulation environments are generated with varying

number of intersection points to effectively simulate packet

Figure 2 – A snapshot of a partial area of the simulationenvironment provided by NCTUns

288289



delays with the number of intersection points. Also for a

fixed number of intersection points, the distances between

the source and destination are varied to get various delays

associated with various distances. To evaluate the

performance on different data transmission density, we vary

the data sending rate (CBR rate) from 0.1 to 1 packet per

second. The performance of the protocols is measured by

the packet delivery delay and data delivery ratio.

A. Packet Delivery Delay

In this section, we compare the data delivery delay for

moving vehicles to fixed sites using VD-4 and H-VADD.

In the first simulation, the packet delivery delay is plotted

against the Euclidian distance between the source and

destination keeping the number of intersections constant.

The plot is shown in the Figure 3.

As can be very easily inferred, the packet delivery delay

increases with the increasing Euclidian distance between

source and destination. This is because once the Euclidian

distance is increased, the paths leading to the destination

must be increased in length to connect the source and

destination. Thus, on an average the length of path between

two intersection points increases. Since VADD considersthat the data packet between two intersection points is

transmitted entirely wirelessly or carried by a vehicle, the

result obtained can be further optimized. This is done by the

VD4 model, which considers the delay to be a sum of

wireless transmission delay and carry delay at the same

time.

Next, the packet delivery delay is plotted against the

number of intersection points keeping the Euclidian distance

between the source and destination as same. The results are

summarized in the Figure 4. As is evident, the packet

delivery delay increases with the number of intersection

points because a fixed number of vehicles are divided into a

large number of paths thus reducing the number of vehicles

on each path, which leads to reduced path density. H-VADD

gives more delay than VD4 protocol because the probability

of choosing a non optimal path increases with each

intersection point. Since intersection points become more,

data delivery delay is increased.

B. The Data Delivery Ratio

In this section we discuss the performance of VD4

protocol as compared to DSR, GPSR, Epidemic routing andH-VADD protocol. The graph shown in the Figure 5 depicts

the relationship between the data rate and data delivery ratio

for the above mentioned protocols. This was obtained as a

result of VD4 simulation with 180 nodes.

Figure 3 – Packet delivery delay versus enclidian distance

keeping number of intersection points constant

Figure 4 – Plot of Packet Delivery delay versus

Number of Intersections

Figure 5 – Plot of Data Delivery Ratio versus Data

Rate

289290



As depicted by the low value of the data delivery delay,

the DSR routing protocol is not suitable for sparsely

connected vehicular networks. GPSR cannot make use of

traffic patterns and the geographical approach used leads to

void areas with a few vehicles passing by. This makes its

delivery ratio poor when the vehicle density is low. With

epidemic routing, when we increase the data rate, the data

delivery ratio goes on decreasing because of the MAC layer

collisions, which increase due to increase in the network

traffic. H-VADD shows a considerably optimal

performance. It reduces geographical forwarding distance

and does not have routing loops. VD4 shows similar

performance to H-VADD. It shows better performance that

H-VADD on lower densities when the roads are not

saturated and the packets need to be transmitted both

wirelessly and carried. Once roads are saturated (at a higher

data rate) almost all transmission is done wirelessly and

delivery ratio increases since packet dropping is reduceddue to carrying of vehicles.

VII. CONCLUSION AND FUTURE WORK

VANETs have been estimated to have tremendous use in

the forthcoming years. The paper discussed the novel VD4

protocol, which adopts the carry and forward technique in

which a vehicle transfers its data to another vehicle if it is

moving towards the destination when it reaches its vicinity.

The VD4 delay model defines the delay associated with a

path as a function of fraction of distance that the packet can

cover when transmitted wirelessly as well as the fraction of

distance in which it has to be carried by the vehicle. Thesimulation was performed in NCTUns 6.0 and the results

obtained thereafter clearly show advantage of the VD4

protocol over other protocols.

As a future work, implementation using real vehicular Ad

hoc network is required to evaluate the protocol in real

world application.

ACKNOWLEGEMENT

Finally, we would like to acknowledge the technical and

financial support provided to by ABV-Indian Institute of

Information Technology and Management Gwalior.

R EFERENCES

[1] H. Wu, R. Fujimoto, R. Guensler, and M. Hunter “MDDV: aMobility-centric Data Dissemination Algorithm for Vehicular

Networks” in ACM VANET, 2004, pp. 47-56.

[2] G. H. Mohimani, F. Ashtiani, A. Javanmard, and M. Hamdi,"Mobility Modelling, Spatial Traffic Distribution, and Probability of Connectivity for Sparse and Dense Vehicular AdHoc Networks" inIEEE Trans. Vehicular Technology, 2008, pp. 1998-2007.

[3] M. K. Mehmet Ali and M. Khabazian, "A performance Modeling of Connectivity in Ad Hoc Networks", in Proc. WCNC'06, 2006, pp.361 - 366.

[4] R.Verdone, "Multi-Hop R-Aloha for Inter-Vehicle Communication atMillimeter Waves", in IEEE Transactions on Vehicular Technology,1997, Vol. 46, No. 4, pp 992-1005.

[5] X. Yang, J. Liu, F. Zhao, and N. Vaidya, "A Vehicle-to-VehicleCommunication Protocol for Cooperative Collision Warning,", in

Proc. MobiQuitous, 2004, pp. 114-123.[6] J. Zhao and G. Cao,"VADD: Vehicle-assisted data delivery in

vehicular ad hoc networks", in Proc. IEEE INFOCOM'06, Apr.2006, pp. 1-12.

[7] F. Li and Y. Wang, "Routing in Vehicular Ad Hoc Networks: ASurvey,", in IEEE Vehicular Technology Magazine, 2007, Vol.2,

NO.2, pp 12-22.

[8] Q. Xu, T. Mark, J. Ko, and R. Sengupta, "Vehicle-to-Vehicle SafetyMessaging in DSRC," in Proceedings of the 1st ACM internationalworkshop on Vehicular ad hoc networks, 2004, pp. 19-28.

[9] J. Yin, T. Elibatt, G. Yeung, B. Ryu, S. Hebermas, H. Krishnan, andT. Talty, "Performance Evaluation of Safety Appications over DSRCVehicular Ad Hoc Networks", in Proceedings of the 1st ACMinternational workshop on VANETs, 2004, pp. 1-9.

[10] S. Ghandeharizadeh, S. Kapadia, and B. Krishnamachari, "PAVAN:A Policy framework for content availability in vehicular ad-hoc-

networks," in Proceedings of the 1st ACM international workshopon Vehicular ad hoc networks, 2004.

[11] Y. Zhang, J. Zhao, and G. Cao, "On Scheduling Vehicle-RoadsideData Access", in Proceedings of the fourth ACM internationalworkshop on VANET, 2007, pp. 9-18.

[12] D. Davis, A. Fagg, and B. Levine, "Wearable Computers as PacketTransport Mechanisms in Highly-Partitioned Ad-Hoc Networks", inInternational Synopsium on Wearable Computing, 2001, pp. 141.

[13] J. Zhao, Y. Zhang, and G. Cao, "Data Pouring and Buffering on theRoad: A New Data Dissemination Paradigm for Vehicular Ad Hoc

Networks," in IEEE Transaction on Vehicular Technology, 2007,Vol. 56, No. 6.

[14] D. Johnson and D. Maltz, "Dynamic Source Routing in Ad HocWireless Networks", in Mobile Computing, Kluwer, 1996, pp. 153-181.

[15]

A. Vahdat and D. Becker, "Epidemic routing for partially connectedad hoc networks", in Technical Report CS-200006, 2000.

[16] B. Karp and H. T. Kung, "GPSR: Greedy perimeter stateless routingfor wireless networks", in Proceedings of ACM MOBICOM-00,2000, pp. 243-254.

290291

Vehicular Density-Dependent Data

Documents