K.PREMA KUMAR et al.

/ (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES

Vol No. 6, Issue No. 1, 124 - 132

A Novel Dangerous Vehicle Detection Protocol For

Safety Application
K.PREMA KUMAR B.M.KUSUMA KUMARI DR.K.SRI RAMAKRISHNA M.KANTI KIRAN
Student M.Tech Lecturer Professor & Head of Department Lecturer
Department TIFAC-CORE in TELEMATICS
V.R. Siddhartha Engineering College ,
Vijayawada, A.P-520007
premakumarkumbha@gmail.com bhupathimercy@gmail.com srk_kalva@yahoo.com kantikiran81@gmail.com

Abstract:- Recently, considerable research has been focused on applying ad hoc wireless networking technology to on-the-
move vehicles. In this paper, we focus on distributed detection of dangerous vehicles on roads and highways using the
dangerous vehicle-detection protocol (DVDP) to detect drivers who violate the permitted speed limit. In DVDP, each vehicle
collects surrounding vehicles identifications (IDs) and propagates warning information (including its position, speed, time,

T
and collected IDs). This information is then forwarded hop-by-hop using ad hoc communications. A vehicle that receives this
information will start to observe its surrounding vehicles. If surrounding vehicles are identified in the received warning
information, it will estimate the speed of such vehicles. If the estimated speed exceeds the permitted speed, such vehicles are
then marked as “suspected vehicles,” and the updated warning information is further propagated. By repeating this process,
ES
the suspected vehicle (where other vehicles observe the speed violation) will ultimately be marked as a “dangerous vehicle.”
This judgment is then further propagated to warn others and to inform the traffic police. We evaluated the performance of
DVDP using a simulator that performs both macroscopic and microscopic traffic simulation, taking into account realistic lane
and speed models, mobility, position, and location errors. Simulation results revealed that DVDP’s detection probability is
greater than 80% when vehicle density is above 40 vehicles/min. When vehicle density is low, deployment of relay points can
help to further improve detection probability. In addition, by utilizing vehicles in opposite lanes, detection probability can be
further improved. This inter-vehicular communication protocol i.e., dangerous vehicle-detection protocol useful for to collect
the fine from the driver who exceeds the permitted speed on highways and to warn other drivers from accidents.
A

Keywords : Mobile ad-hoc networks , mobile communication , vehicular mobility and vehicle-to-vehicle communication.

I . INTRODUCTION
IJ

Vehicular Ad Hoc Networks (VANETs) have recently
attracted the attention of researchers , automotive
industry, and governments , for their potential of improving
safety of the surrounding environment through infrastructure
less ,vehicle to-vehicle(V2V) wireless communications The
vision for VANETs includes applications such as route
planning, road safety, e-commerce , entertainment in for
VANETs includes applications such as route planning, road
safety, e-commerce , entertainment in vehicles
cooperative driver assistance, sharing traffic and road
conditions for smooth traffic flow, user interactions,
information services, etc. VANET raises several interesting
issues in regard to media access control (MAC), data
aggregation, data validation , data dissemination , routing ,
network congestion , performance analysis, privacy and
security[5] .In VANETS communications between nodes are
direct ,i.e. without relying on any dedicated infrastructure
,contrary to earlier vehicular networks Although it is self-
organized and easily to deploy, the infrastructure less
vehicular network introduces many challenges that should
be tackled before real implementations. For instance, to allow
communications between nodes (vehicles) which are out of
[2], the power range of each other some other intermediaries
should act as routers, to remedy the lack of dedicated routers.
Thus, a distributed routing protocol needs to be employed. In
the European Union Seventh Framework Program me
SAFESPOT project [11],[12] and the Japanese Advanced
Road Transportation Systems (ARTS) project, future

ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 124

 K.PREMA KUMAR et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
intelligent transportation system. technologies based on
vehicle-to-vehicle (V2V) and vehicle to- roadside (V2I) [9]
[10] communications are actively being studied .In fact, V2V
is considered useful for localized emergency and respond
cases, while V2I is viewed as the long-haul pipe for
noncritical information push and pull. In the ARTS project,
there are several research projects on cooperatively detecting
vehicles from approaching ramps, invisible corners,
intersections, vehicles in the wrong lane, vehicles travelling at
extremely dangerous speeds, and vehicles wrongly stopping or
running on a motorway shoulder. In this paper, we focus on
the use of ad- hoc communications to detect dangerous
vehicles on roads and highways [6],[7],[8] . Reckless and
careless driving can endanger the lives of other drivers and
passengers on roads and highways. Increasingly, many
accidents can be avoided if such dangerous drivers are
detected early and other drivers are warned [14][15]. In most
roads nowadays, cameras and speed detectors are used to
ES
monitor and identify drivers who had exceeded the permitted
speed limit on roads and highways. This is the primitive
approach, and there are limitations. If the drivers slow down
before the speed detectors, they will not be detected, even
though they had exceeded the permitted speed earlier.
Additionally, bad weather can distort the quality of image
captured by cameras. Currently, there are various
interpretations and definitions of dangerous drivers. Reckless
A
drivers are those who are negligent. For example, they do not
give signals when overtaking other vehicles, or they follow
too closely to vehicles ahead of them. Fatigue drivers are
those who are overpowered by physical and mental stress.
They fell asleep while driving, creating a dangerous situation
IJ
on roads and highways, resulting in accidents and Collisions
19] . Aggressive drivers are those who drive and overtake
others dangerously, often speeding, and do not give warning
or signals to others. They create catastrophic accidents, often
involving fatal injuries and loss of lives. In this paper, we
focus our research on the detection of drivers who violate the
speed limits on roads and highways [16],[18] . We used the
dangerous-vehicle-detection protocol (DVDP) to detect
drivers who violate the perm permitted speed limit .Here ,we
assume that a kind of next-generation license plate number
system is used, where each vehicle disseminates its own
vehicle identification (ID) in a regular interval. In DVDP,
each vehicle collects surrounding vehicles’ IDs and
ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved.
Vol No. 6, Issue No. 1, 124 - 132
propagates warning information including their position,
speed, time, and collected IDs. When a vehicle receives
warning information, the vehicle calculates the distance
between the current position and the position where warning
information was generated. If the distance is shorter than a
certain amount, the vehicle will forward the warning
information to preceding vehicles as soon as possible.
Otherwise , the vehicle that receives the warning information
starts to observe its surrounding vehicles. When it finds
vehicles whose IDs are included in its received warning
information, it estimates the speed of those vehicles. If the
estimated speed exceeds the permitted speed limit ,those
vehicles are considered as ―
suspected vehicles,‖ and the
updated warning information is then further propagated to
T
vehicles ahead. By repeating this process, the suspected
vehicle where multiple other vehicles witnessed the speed
violation will ultimately be marked as a ―
dangerous vehicle.‖
Fig. 1. Overview of the proposed dangerous-vehicle-detection mechanism
II. DANGEROUS VEHICLE DETECTION PROTOCOL
Current methods of deploying speed detectors at
fixed points are inadequate for detecting certain kinds of
dangerous vehicles. For example, if a dangerous vehicle
exceeding the permitted speed limit knows the locations of
those speed detectors, it might slow down before the speed
detectors, thereby avoiding detection. In this paper, we
propose DVDP to detect dangerous vehicle autonomously. Its
operation is discussed in the following section.
A . Assumption for Proposed Protocol:
Here, we assume a kind of next-generation license
plate system. We assume that each vehicle broadcasts its
Page 125
 K.PREMA KUMAR et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
vehicle ID periodically, and that the vehicle ID’s radio range
is 40 m. It is also assumed that those vehicles have wireless
communication facilities so that they can transmit and receive
data. For inter-vehicular communications, we use the
IEEE802.11 independent basic service set (IBSS) [3], [4]. Our
protocol is an application layer protocol running over user
datagram protocol/ Internet protocol (UDP/IP) and
IEEE802.11. Here, we assume that the wireless
communication range is 200 m and that each vehicle knows its
precise position and time from its global positioning system
(GPS) receiver. Different radios are used for transmitting data
and vehicle IDs. In DVDP , each vehicle collects vehicles’ Ids
disseminated from surrounding vehicles and forwards them to
preceding vehicles as warning information (see Fig. 1). Note
that DVDP can work well even if dangerous vehicles do not
execute DVDP although we assume that all vehicles including
dangerous vehicles broadcast their vehicle ID’s
periodically[1].
ES
B . DVDP Operation
Let Vid denotes the vehicle whose vehicle ID is id.
Warning information Wid of Vid contains the following four
types of information ( Wid = < Oid , Mid, Sid , Did > ): 1)
observing vehicle information Oid; 2) monitoring vehicle
information Mid; 3) suspected vehicle information Sid; and 4)
dangerous vehicle information Did. The observing vehicle
information Oid is represented as 4- tuple ( Vid, tid, pid, dirid
A
) , where pid denotes the position of the vehicle Vid at time
tid, and dirid denotes the direction in which it is moving.
When other vehicles receive the warning information, they
can recognize where the vehicle Vid is running at time tid and
to which direction it is moving from pid and dirid,
IJ
respectively.
The monitoring vehicle information Mid is represented as a
set { TRid } of monitoring vehicles’ moving traces. The
moving trace TRid of each monitored vehicle Vid is
represented as a pair {trid, id } of a list of positions trid and its
vehicle’s ID
id_. Here, trid denotes a list of positions { ( t1, p1,
v1),( t2, p2, v2 ), . . . , ( tk, pk, vk ) }, where pi and vi denote the
estimated position and speed of the monitored vehicle Vid_ at
time ti, respectively. The estimated speed vi is calculated by
comparing its current position and time with its preceding
position and time .
The suspected vehicle information Sid is represented as a
set of suspected vehicles Sid = { TRid, levelid }, where levelid_
ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved.
Vol No. 6, Issue No. 1, 124 - 132
denotes the suspected level of a monitoring vehicle Vid .When
vehicle Vid estimates the speed of each surrounding vehicle
Vid , if it finds that Vid exceeds the permitted speed limit, i.e.,
if the estimated speed vk in TRid exceeds the permitted speed
limit, the vehicle Vid sets {TRid , 1 } into its suspected vehicle
information Sid. Also, if vehicle Vid finds again that Vid in Sid
still exceeds the permitted speed limit, the suspected level is
incremented by one, i.e., it is incremented as { TRid , levelid +
1} .The vehicles in Sid correspond to questionable vehicles,
and they are continuously monitored by other vehicles . If the
suspected level reaches a predefined threshold N , { TRid ,
levelid } is recorded into the dangerous vehicle information
Did. The data structure of dangerous vehicle information Did
is the same as the suspected vehicle information Sid. If
T
preceding vehicles receive the dangerous vehicle information
from vehicles in the rear, they can continuously monitor the
dissemination of vehicle IDs of those vehicles. If they find
such dissemination of vehicle IDs, they can recognize the
approach of dangerous vehicles in advance. Hence, this can
enhance the safety of preceding vehicles.
Fig.2 Dangerous vehicle speed estimation
C. Detection Approach
In DVDP, each vehicle executes the following two
processes in parallel. 1. forwarding process and 2.Observing
process. The forwarding process relays the warning
information sent from rear vehicles to preceding vehicles. It
also relays the warning information of oncoming vehicles so
that they can be used for vehicles running in opposing lanes
.The observing process, however, monitors dissemination of
vehicle IDs from surrounding vehicles and estimates the speed
of surrounding vehicles using the warning information
obtained in the forwarding process. The forwarding process
Page 126
 K.PREMA KUMAR et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 6, Issue No. 1, 124 - 132

has a forwarding list for keeping the warning information to b) If id’ is not included in the observing list, then construct its
be forwarded. Likewise, the observing process has an moving trace << tcur, pcur, vcur> , id’>, and add it to the
observing list for keeping vehicle IDs to be monitored. forwarding list.
1). Forwarding Process: The forwarding process is executed c) If id_ is included in the observing list and if <<t1, p1, v1>,

as follows. <t2, p2, v2>, . . . , <tk, pk, vk>, id’> is recorded as a moving

a) If a vehicle Vid receives a warning information Wid, it trace, then estimate its current speed Vcur as follows:

compares its current position of Vid with the position of the Vcur = d(pk, pcur) − 2R /tcur – tk

sender vehicle (Vid ) recorded in Wid. It also compares the where d(p1, p2) denotes the distance between p1 and p2.

moving direction. d) Let Vlimit denote the permitted speed limit. If Vcur > Vlimit,

b) If Wid is sent from an oncoming vehicle running on the then increment the suspected level lid , and put it on the

opposite lanes, then add Wid into the forwarding list, and go to forwarding list. Let L denote a threshold. If lid > L, then treat

step 5). the corresponding vehicle as a dangerous one, and put it on the

c) If Wid is sent from a front vehicle running in the same forwarding list.

direction, then discard it, and repeat from step 1). e) For a predefined speed Vsend, if Vsend ≤ Vcur ≤ Vlimit, then

T
d) If Wid is sent from a rear vehicle running in the same retain the suspected level lid, and add the corresponding

direction, then do the following. warning information on the forwarding list.

1) For each moving trace <trid’’, id‖> in the monitoring f) Otherwise (Vcur < Vsend), remove the target vehicle Vid from

vehicle information Mid_ = {<trid‖, id‖>} of Wid ,if the the observing list.
ES
distance between the latest position of the moving trace <trid‖, Here, we assume that the observing list is periodically

id‖> and the current position is less than a fixed distance Dmin, updated and old information is discarded. For each

then put the moving trace <trid‖, id‖> into the forwarding list. information whose latest timestamp is older than a fixed

Repeat this process for all moving traces in Mid’. period Top, observation of such vehicles is no longer

2) Else, if the distance is larger than or equal to Dmin, then add necessary, and the old information is discarded. In each

<trid, id> into the observing list. moving trace TRid_ = <{<t1, p1, v1>, <t2, p2, v2>, . . . , <tk, pk,

e) If the forwarding list is not empty, then disseminate its vk>}, Vid’> , a new list <tcur, pcur, vcur> is added only when the

contents as a new warning message, and remove them from suspected level is incremented. Therefore, the maximum
A

the list. length of the list is N + 1, where N is the predefined threshold

The speed check of suspected vehicles will begin after for deciding whether a suspected vehicle is treated as a

the warning information is propagated over the fixed distance dangerous vehicle. If each<ti, pi, vi> is given as 20-B data,

Dmin. When the vehicular distance is short, vehicles driving in and if Vid’ is given as 4-B data, the size of a moving trace
TRid’ is 20 * N + 24 B .For neighboring vehicles running with
IJ

front may encounter the suspected vehicle before they receive

the warning information. Here, if the differential angle similar speeds, their suspected levels are zero. In such a case,

(absolute value) between the moving direction described in the the required data size is 24 B since N = 0. Although the size of

received warning information and the current vehicle moving warning information varies depending on how many vehicles’

direction is less than π/2, both vehicles are considered to be information should be forwarded, in our experiments, the sizes

moving in the same direction. Else, they are said to be moving of the warning information are smaller than 1 kB in most

in opposite directions. cases. The number of transmitted warning information per

second within the wireless communication range of each
2 .Observing Process:Each vehicle Vid executes the following
vehicle is at most 20–50, even for rather congested road
observing process (see Fig. 2).
conditions. This amount is enough for transmitting data in
a) If vehicle Vid finds dissemination of a vehicle ID id’ from its
IEEE 802.11p environments .When estimating the speed of a
surrounding vehicle Vid’ , then note its own current position
surrounding vehicle, we cannot find its exact position.
pcur, speed vcur, and time tcur.

ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 127

 K.PREMA KUMAR et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
In Fig. 2, if R is the wireless communication range, the real
position of an observed vehicle might be at most R meters
.away from the position of its monitoring vehicle. Therefore, if
the distance between two monitoring vehicles is X meters, the
actual distance X_ between two positions of vehicle VX
satisfies X − 2R ≤ X_ ≤ X + 2R. Therefore, we use the
minimum distance X − 2R as the distance between two
positions of vehicle VX when estimating its speed. Since X −
2R is used, the estimated speed is slower than the actual
speed. This avoids cases where non dangerous vehicles are
detected as dangerous vehicles.
The warning information is also transmitted when the
estimated speed is faster than a fixed speed Vsend, even if the
speed is less than the permitted speed limit Vlimit. If the
warning information is not sent when the estimated speed is
lower than Vlimit, dangerous vehicles might not be detected if
they temporarily slow down. By relaying warning information
of vehicles whose speeds are faster than Vsend, we avoid
ES
misdetection dangerous vehicles. Hence, we refer Vsend as the
forwarding speed.
III . REALISTIC VEHICULAR MOBILITY
In the evaluation of inter-vehicular communication, it is
desirable that realistic vehicular traffic flows are considered. If
inter-vehicular communication is used for estimating the
duration of a traffic jam, one can assume that vehicles in the
A
jam form a line with similar distances, i.e., they have uniform
arrangement. If inter-vehicular communication is used for
avoiding collisions at intersections, one can assume that a
message informing the approach of a vehicle can reach
vehicles on its crossing road by at most a few hops. In such
IJ
cases , performance of inter-vehicular communication for real
traffic might be similar to simulation results based on artificial
macroscopic traffic flow models. However, to identify moving
traces of dangerous vehicles and estimate when and how
dangerous vehicles exceed the permitted speed using inter-
vehicular communication, one must carefully reproduce real
traffic flows on highways for considerable time periods. This
requires a realistic vehicular mobility model. In this section,
we consider vehicular mobility by taking into account realistic
speed model and lane selection/change model. We have
extracted specific characteristics of real vehicular mobility
from several vehicular positions on Google Earth , and we
have shown that our vehicular mobility model can
ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved.
Vol No. 6, Issue No. 1, 124 - 132
successfully reproduce vehicular density and fluctuations of
real traffic flows.
A. Speed Model: According to the theory of traffic
engineering, it is known that, in general, each vehicle’s speed
V can be modeled by the following Green shields’ function
[17]:
V = VMAX ×(1 – K/KMAX ) - (1)
V speed of the target vehicle;
K density of vehicles around the target vehicle;
VMAX maximum speed at which vehicles can run in case that
K = 0;
KMAX saturated density of vehicles when their speeds become
T
0.
Here, we assume that the ratio K/KMAX is roughly proportional
to the ratio DMIN/D .Then, we arrive at
V = VMAX ×(1 − DMIN /D) - (2)
where D is the distance to the preceding vehicle, and DMIN is
the minimum distance between two vehicles.
IV.SIMULATION ENVIRONMENT
A. Simulation environment:
we evaluated the performance of our proposed DVDP on
metropolitan highways. The common simulation condition is
given as follows. We assume that the legally allowed highway
speed is 100 km/h and that each normal vehicular speed is set
at 100 ±5 km/h . We vary the vehicular density from 5 to 55
vehicles/min. Each highway has two or three lanes, and there
is almost no traffic congestion for the given vehicular density
.Here, vehicles with an average speed exceeding 130 km/h are
classified as dangerous vehicles. The number of dangerous
vehicles is set to 9.For inter-vehicular communications, we
use the IEEE802.11 IBSS [7], [8]. Our protocol is an
application layer protocol running over UDP/IP and
IEEE802.11. The warning information dissemination range is
200 m, and the vehicle ID’s radio range is 40 m. Different
radios are used for sending data and vehicle IDs. For wireless
communications, it is known that the probability of successful
reception of each packet depends on the inter-vehicular
distance and that it decreases suddenly when the distance
becomes larger [20]. Here, we approximate that if packet
Page 128
 K.PREMA KUMAR et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
collision does not occur, the correct reception probability of
the warning information, which depends on the inter-vehicular
distance, follows the expression 1 − (x/D)2, where x denotes
the inter-vehicular distance, and D denotes the data
dissemination range. If packet collision occurs, then we say
that collided packets are lost . We assume all vehicle IDs
from surrounding vehicles can be received at the same
probability of 60% for simplicity of discussion. Here, every
vehicle emits its vehicle ID every second. Since the relative
.
ES
Fig.3. Suspected level versus Dmin
A
IJ
Fig .5. Detection probability with location and time error
ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved.
Vol No. 6, Issue No. 1, 124 - 132
speed between the normal vehicles and dangerous ones is at
most 100 km/h (28 m/s), and it is less than the radio range of
vehicle IDs (40 m), each vehicle can receive the vehicle ID of
every passing vehicle with a high probability, even if the
passing vehicle’s speed is very fast .In our simulation, we will
evaluate the performance of DVDP in scenarios with and
without relay points. Our simulation results are discussed
below
T
Fig .4. Correlation between Dmin and detection probability
Fig. 6. Detection probability with the help of vehicles in opposite lanes
Page 129
 K.PREMA KUMAR et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Fig.7. Detection probability versus forwarding speed.
B. Dmin Versus Suspected Level Versus Detection Probability
ES
In the forwarding process, when warning
information is sent from a rear vehicle to its preceding vehicle,
if the distance between the two vehicles is shorter than Dmin,
the preceding vehicle just forwards the received warning
information to its preceding Vehicle. The speed check is
carried out only if the distance between the two vehicles
exceeds Dmin. Here, we have evaluated the influence of Dmin
on suspected levels of dangerous vehicles. The two evaluation
A
parameters are listed as follows:
1) Vehicle density = 31.7 vehicles passing in a minute;
2) Dangerous vehicles speed = 150 km/h.
Fig. 3 shows the increase in suspected levels for different
IJ
values of Dmin. As Dmin decreases, the suspected level
increases quickly . As Dmin increases ,the suspected level
increases with a lower gradient. While a shorter Dmin increases
the suspected levels quickly, too short a distance can result in
the failure of detecting dangerous vehicles.
C. Position and Time Errors:
In our DVDP, the speed of each observed vehicle is
checked when it is recognized that the vehicle moves Dmin
meters. As shown in Fig.2. Suspected Level versus Dmin
i
meters, the actual moving distance X of each observed vehicle
might be 2+R meters longer than the estimated distance X. If
it takes T seconds to move Dmin meters, the actual speed might
be (Dmin + 2 * R)/T , although the estimated speed is Dmin/T. If
Dmin is very short, the difference between the actual speed Vi
ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved.
Vol No. 6, Issue No. 1, 124 - 132
and the estimated speed V becomes large. This makes the
detection probability low since the estimated speed is rather
slower than the actual speed. On the other hand, if Dmin is very
long, the warning information is often lost due to the absence
of forwarding vehicles. As shown in Fig. 4, in our experiments
,if Dmin is 300 m, the detection probability is less than 40%. If
it increases, i.e., if it is between 500 and 800 m, the detection
probability becomes 60%. If it exceeds 1000 m, the detection
probability decreases. From these results, we choose Dmin
=800 m .We discuss here the error between the estimated
speed and the actual speed caused by position and time errors.
Here, we assume that the position error is −10 < dx < 10 m
and that the time error of GPS is −0.5 < tx < 0.5 s. The
T
position and time errors are present in both front- and rear-
observing vehicles. Hence, the estimated speed V _ is
i
calculated as V = X + 2dx − 2R/T + 2tx. Here, X and T
denote the exact location and time. By considering the range
of dx and tx from the assumption, the range of V _ is (X − 20
− 2R/T + 1) < Vi < (X + 20 − 2R/T − 1). Fig.5 denotes the
detection probability of dangerous vehicles with location and
time errors. This evaluation was done with the dangerous
vehicle’s speed at 150 km/h and Dmin at 800 m. As shown,
there is no major difference between the detection
probabilities with and without errors. Hence, our proposed
DVDP can work well even if vehicles do not have exact
location and time information.
D. Influences by Vehicles From Opposite Lanes :
So far, we have focused on vehicles running in the same
direction. Here, we utilize vehicles in the opposite lanes to
relay warning information. This can further improve detection
probability . Fig. 6.denotes the detection probability for the
following four cases.
1) Observing and forwarding vehicles are the only vehicles
driving in the same direction with the dangerous vehicles.
2) Observing vehicles are vehicles driving in the same
direction with the dangerous vehicles and forwarding vehicles
are vehicles driving in both lane directions.
3) The role of vehicles is the same as the first case, but
additional relay points are set per 1 km by the roadside.
4) The role of vehicles is the same as the second case, but
relay points are set per 1 km by the roadside.
As shown in Fig.6, with vehicles running in the same lane and
direction, as vehicle density increases, detection probability
Page 130
 K.PREMA KUMAR et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 6, Issue No. 1, 124 - 132

also increases, reaching above 80% beyond 40 vehicles/min.
However, with the inclusion of vehicles from opposite lanes ,
detection probability is greatly improved
.E . Detection Probability Versus forwarding Speed:
Next, we focus on the forwarding speed Vsend. If a
dangerous vehicle slows down temporarily, the estimated
speed might temporally become less than the speed limit Vlimit
(even if the average speed exceeds the speed limit Vlimit).
Therefore, to detect such a case, we introduce the forwarding
speed Vsend , whose value is less than the speed limit Vlimit,
and the warning information is forwarded when the estimated
speed exceeds Vsend .When the estimated speed is less than
Vlimit and greater than Vsend , the suspecting level of the
warning information is not increased, and the warning
models, speed of dangerous vehicles, forwarding speed, relay
points, and vehicles in opposite lanes . Simulation results
revealed that detection probability for realistic vehicular
mobility is less than that of the macroscopic case. Increasing
forwarding speed reduces detection probability . We
observed that the deployment of relay points increases
detection probability. Furthermore, the higher the average
speed of dangerous vehicles, the higher the detection
probability. Exploiting vehicles in the opposite lanes to relay
warning information can further improve detection
probability. From our results, it is clear that distributed
detection of dangerous vehicles is possible using ad hoc
sensing ,communications ,and relay points.

T
information must continue to be forwarded. We evaluate VI.REFERENCES:
detection probability by varying the forwarding speed Vsend. In
Fig.7, we show the detection probability under various [1]. S. Panwai and H. Dia, ―
Comparative evaluation of microscopic car
following behaviour,‖ IEEE Trans. Intell. Transp. Syst., vol. 6, no.
forwarding speed Vsend: 100–125 km/h. In this evaluation, the
3, , Sep. 2005
ES
average speed of dangerous vehicles is 134 km/h, and the
[2]. Hartenstein, B. Bochow, A. Ebner, M. Lott, M. Radimirsch, and D.
maximum speed is 150 km/h. The vehicular density is 32 Vollmer, ―
Position-aware ad-hoc wireless networks for inter-
vehicles passing in a minute . Fig. 7 shows that the slower the vehicle communications: The Fleet Net project,‖ in Proc. 2nd ACM
Int. Symp Mobi Hoc, 2001,
forwarding speed, the higher the detection probability. This is
[3]. Wireless LAN Medium Access Control (MAC) and Physical Layer
because when the forwarding speed is slower, the number of
(PHY) Specifications, IEEE802.11 Std., 1999. ISO/IEC8802-
warning information transmitted is increased, and dangerous 11:1999.
vehicles can be readily detected. When the forwarding speed [4]. Supplement to 802.11-1999, Wireless LAN MAC and PHY
Specifications: Higher Speed Physical Layer (PHY) Extension in
is 100 km/h, the detection probability becomes low because
A

the 2.4 GHz Band, IEEE802.11b Std., 1999. ISO/IEC8802-

the warning information from normal vehicles is also sent, and
11:1999.
the number of the warning information becomes too large. [5]. A. Khelil, C. Becker, J. Tian, and K. Rothermel, An epidemic
Therefore, the warning information of dangerous vehicles model for information diffusion in MANETs,‖ in Proc. 5th ACM
Int. Workshop MS Wi M, 2002,.
cannot be propagated effectively. .
[6]. .G. Korkmaz, E. Ekici, F. Ozguner, and U.ozguner, Urban multi-
IJ

hop broadcast protocol for inter-vehicle communication systems,‖

V.CONCLUSION in Proc. 1st ACM Int. Workshop VANET, 2004,
In this paper, we have proposed the DVDP an inter- [7]. R. Mangharam, D. Weller, R. Rajkumar, P. Mudalige, and F.
Bai, ―
GrooveNet: A hybrid simulator for vehicle- to-vehicle
vehicular communication protocol for distributed detection of
networks,‖ in Proc. 2nd Int. Workshop V2VCOM, 2006. CD-
dangerous vehicles. DVDP uses ad hoc mobile
ROM.
communications to propagate warning information to [8]. R. Miller and Q. Huang, ―
An adaptive peer-to-peer collision
preceding vehicles so that dangerous vehicles can be warning system,‖ in Proc. IEEE VTC, 2002
[9]. T. Nadeem, S. Dashtinezhad, C. Liao, and L. Iftode, Traffic
monitored and identified over time and space. I shall
view: A scalable traffic monitoring system,‖ in Proc. 5th IEEE
implement a realistic simulator that perform both
Int. Conf. MDM, 2004.
macroscopic and microscopic simulation. We had also taken [10]. K. Nakanishi, T. Umedu, T. igashino, H. Mori, and H.
into account appropriate speed and lane models, along with Kitaoka, ―
Synthesizing realistic vehicular mobility for precise
simulation of inter-vehicle communication,‖ in Proc. 2nd IEEE
considerations for time and position errors. We evaluated the
Workshop AutoNet, 2007. CD- ROM.
performance of DVDP by examining detection probability
over various parameters, such as vehicle density, mobility

ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 131

 K.PREMA KUMAR et al. / (IJAEST) INTERNATIONAL JOURNAL OF ADVANCED ENGINEERING SCIENCES AND TECHNOLOGIES
Vol No. 6, Issue No. 1, 124 - 132

[11]. S. Ni, Y. Tseng, Y. Chen, and J. Sheu, ―

The broadcast storm
problem in a mobile ad hoc network,‖ in Proc.5th ACM/IEEE Int.
Conf. MobiCom, 1999,.
[12]. An Inter-vehicular-Communication Protocol for Distributed
Detection of Dangerous Vehicles Takaakimedu, Member, IEEE,
Kumiko Isu, Member, IEEE ,Teruo Higashino, Senior Member, C.
K. Toh, Fellow ,IEEE transactions on vehicular technology, vol. 59,
no. 2, February 2010.

[13]. S. Panwai and H. Dia, ―

Comparative evaluation of
microscopic car following behaviour,‖ IEEE Trans. Intell. Transp.
Syst., vol. 6, no. 3, pp. 314–325, Sep. 2005.
[14]. M. Saito, J. Tsukamoto, T. Umedu, and T. Higashino, ―
Design
and evaluation of inter-vehicle dissemination protocol for
propagation of preceding traffic information,‖ IEEE Trans. Intell.
Transp. Syst., vol.8, no. 3, Sep. 2007.
[15]. C. Schwingenschlogl and T. Kosch, ―
Geocast enhancements
of AODV for vehicular networks,‖ ACM SIGMOBILE Mobile

T
Comput. Commun. Rev., vol. 6, no. 3, pp. 96–97, Jul. 2002.
[16]. in Proc. 5th World Congr.Intell. Transp. Syst., 1998, vol. 4,
M. Torrent-Moreno, D. Jiang, and H. Hartenstein, ―
Broadcast
reception rates and effects of priority access in 802.11-based
vehicular ad-hoc networks,‖ in Proc. 1st ACM Int. Workshop
ES
VANET, 2004.U.S. Dept. Transp., Fed.Highway Admin., Traffic
[17]. Flow Theory—A State of the Art Report.
[Online].available:http://www.tfhrc.gov/its/tft/tft.ht
[18]. B. Xu, A. Ouksel, and O.Wolfson, ―
Opportunistic resource
exchange in inter-vehicle ad-hoc networks,‖ in proc. 5th IEEE Int.
Conf. MDM, 2004
[19]. J. Yoon, M. Liu, and B. Noble, ―
Sound mobility models,‖ in
Proc. 9th ACM/IEEE Int. Conf. MobiCom. .
A
IJ

ISSN: 2230-7818 @ 2011 http://www.ijaest.iserp.org. All rights Reserved. Page 132