VANET plays an important rule nowadays aiming for safer roads. To increase the road safety VANET allows cars inside the network to exchange warning messages to warn other vehicles about the hazards when it happen. When any danger is detected an emergency message need to be transmitted for a wide range, and for large number of vehicles to let other vehicles aware of the danger before reaching it, this critical information must be sent with high probability, reliability and efficiency.
In order to disseminate the emergency information, vehicle inserts the information into a message and sends this message in broadcast fashion, broadcasting the message to the current network may not guarantee to the message to reach all the vehicle that may get benefit from it due to network considerations or problems like fading or channel jam.
The probability of reception can reach 99% in short distances and can be as
low as 20% at half of the communication range (Torrent-Moreno
et al., 2004). Vehicles in longer distance seem to be interested
in receiving the emergency message as it will reach the danger area after short
time due to the high speed of vehicles moving in VANET network, so we need a
technique to insure the reception of the message for vehicles in long distances.
In this study we concerned with design a new protocol that will increase the percentage of reception in higher distances which will help the vehicles in the network to avoid the danger.
ANALYSIS OF RELEVANT RESEARCH AREA
Many studies have introduced the idea of how to increase the reception of the
emergency message. Mittag et al. (2009) proposed
a comparison between sending the information in single-hop transmission at high
transmission power and multi-hop transmission at lower transmission power, the
comparison aimed to reduce the congestion (load) on the channel caused by periodic
safety messages (Beacons). To enable data to reach higher distances authors
suggested sending the data in multi-hop at low power, which will decrease the
congestion in the network and enable the data to reach higher distances.
Fubler et al. (2003) proposed the Contention-Based
Forwarding (CBF) protocol where a node sends a packet as a broadcast message
to all of its neighbors. On receiving the packet, neighboring nodes will contend
for forwarding the packet by means of waiting for a period of time called contention
time. The node having the maximum progress to the destination will have the
shortest contention time and will first rebroadcast the packet. If other nodes
receive the rebroadcast message, they will stop their contention and delete
the previously received message. This protocol mainly proposed for forwarding
the periodic safety message (Beacons). The problem of this protocol that there
should be a management technique to manage the contention for all the neighboring
vehicles, and there is a chance that the nearest vehicle to the sender may not
hear the rebroadcast of another vehicle, in this case this vehicle will rebroadcast
the message (hidden node problem (Tobagi and Kleinrock,
1975; Khan et al., 2008) causing to broadcast
storm problem and making the protocol useless.
Another forwarding scheme proposed by Korkmaz et al.
(2004) aiming to maximize the message progress, broadcast storm, hidden
node and reliability problems. The protocol assigns the duty of forwarding and
acknowledging the broadcast packets to only one vehicle by dividing the road
portion inside the transmission range into segments and choosing the vehicle
in the furthest non-empty segment without prior topology information. The weakness
of this protocol is the high delay as sender and receiver must make request
to broadcast (RTB) and clear to broadcast (CTB) handshake to select the forwarder
before sending the message and this procedure may occur more than one time to
verify the next forwarder.
Mobility prediction method presented by Meng et al.
(2008) is not suitable in high mobile network like VANET.
Moreno (2007) and Torrent-Moreno
et al. (2009) proposed position-based message forwarding strategy
in order to disseminate time-critical safety information, the strategy depends
on sending the message in broadcast fashion. All the vehicle receivers are a
potential forwarder. Before sending the emergency message, the next forwarder
will be selected as the furthest vehicle in the dissemination area. The problem
of this strategy that the forwarder itself may not receive the message to forward
it due to channel fading as it is very far from the source creating a hidden
The proposed Emergency Message Dissemination for Vehicular (EMDV) (Torrent-Moreno,
2007) protocol tries to maximize the reception of the emergency message
by enabling the furthest vehicle within the transmission range to make the rebroadcasting
of the emergency message, choosing one forwarder vehicle is not appropriate
in high mobile network like VANET as the position is always changing, and the
receiver vehicle may become out of the range when sending the message or simply
the receiver cant receive the message because of the channel problems
like jam or denial of service.
Basic idea assumptions: In VANET each vehicle sends to all neighbors
a beacon message every 100 ms (White Paper, 2005), helping
them to detect and avoid potential dangers, these messages considered as life-critical
messages and contain ID, position, direction, speed, time stamp and beacon interval
(Abuelela and Olariu, 2009; Samara
et al., 2010). Beacons should be received by all the neighboring
vehicles with high probability and reliability, we also assume that each vehicle
equipped with a GPS device to retain the current position (Wang
et al., 2008).
We assume that when a vehicle detecting a danger it will issue an emergency
message to warn other vehicles about it. When sending a message, vehicles until
the back end of the sender transmission range must receive it, i.e., all the
vehicles in the opposite direction of the sender movement located in the coverage.
It must be noted that covering the whole area does not guarantee that all the
vehicles will receive the message due to channel collisions and fading effects,
it must be insured that the emergency message percentage of reception for the
network vehicles must be as high as possible.
In order to cover a wider area for message reception some neighboring vehicle is a potential forwarder, each forwarder will contend by means of waiting for a period of time called contention time before making the forwarding.
Preparing to send: Every beacon received by a vehicle provides important information about the sender status, these information could be utilized to form a rich and real time image about current network topology which will help network vehicles to communicate with each other on better way, also it helps to know about the hazards quickly when it happen.
When a vehicle has a problem or detect problem it analyze it to know its kind, is it a life critical or not? the life critical (Safety of Life) messages will be given the highest priority to be processed and sent before any other kind of messages, for this purpose we propose to categorize any emergency message before sending it, to make it easier to the receiver vehicle to recognize the received message importance, Table 1 which contains the code for each category, for example when a vehicle receives two messages contains the category 1 and 5 it will process the message that contains category 1 first, as it contains more important and critical information.
After assigning the message code, the sender should add data to the message like coordination of the hazard or what the receiver should do; we will not discuss this point as it is out of the scope of the study.
The proposed structure of the emergency message sent is given in Fig. 1.
Sending the message in single hop will enable it to reach for number of vehicles
and for a limited distance up to 1000 m for the best cases (White
Paper, 2005) but this number must be increased to more vehicles aware of
the danger before reaching it.
||Emergency message illustration. Sen ID: Sender id, Code: Message
Code, TS: Time Stamp, Msg ID: Message ID, Data: Data sent, CId: Forwarder
Candidate ID, MinP: Minimum boundary, MaxP: Maximum boundary, Hop: Number
of message hops
|| Emergency message classification
Choosing the next candidate forwarder is a process starts by gathering the information gained from beacons received from neighbors, this information is inserted and ordered into Table 2 NT, NT contains information about neighbors as like: ID, Position, Speed, Heading (direction is the same or deferent), Table 2 provides a fresh information about neighbors and helps to make a quick decision to choose the next forwarder, example on NT in Table 2 based on Fig. 2 scenario.
The sender vehicle chooses the furthest vehicle and assigns it to be the candidate forwarder, forwarding the emergency message will be our option to increase the probability of the reception as the forwarder signal will reach more vehicles on the road and arrive for further distances.
As we mentioned before, choosing one forwarder is not appropriate in high mobile network like VANET, as the forwarder may not receive the emergency message, to solve this problem we propose to divide the network to several segments, vehicles inside the last none empty segment (furthest segment from the sender) will contend for a period of time and check if the candidate forwarder rebroadcasted the emergency message or not, if no one made the rebroadcast, vehicles located in the furthest segment will forward the message.
Vehicles in the last segment, which should be the furthest one, will make the forwarding of the emergency message in case the forwarder fails to receive and forward the message.
Dividing the network to several segments is proposed by Korkmaz
et al. (2004), Zhou et al. (2009),
Dai et al. (2009) and Fasolo
et al. (2006). As in study of Korkmaz et al.
(2004) the sender transmission range been divided into 10 segments to make
it easier on the sender to discover the last vehicle in none empty segment which
will be selected as the next forwarder. For present study, this distance will
be the distance between the sender and the forwarder, as the forwarder which
is the furthest neighbor maybe away by 900 m not 1000, so the distance area
after 900 m must not be considered for better performance, while the distance
in study of Korkmaz et al. (2004) is a fixed
distance which equals to 1000 m.
|| Neighbor Table based on fig. 2 scenario
||Emergency message sending and transmission range
To compute the boundaries of the last segment the sender implements the following equations:
Dis is the distance between the sender and the forwarder, SenPos is the sender position, ForPos is the forwarder position, MaxB is the highest boundary (border) for the last segment, MinB is the minimum boundaries (Borders) from where the segment starts, Nmax is the maximum number of the segments, Dif is the length of the segment.
This means that the vehicles located in the area between MinB and MaxB from the sender will be considered as potential forwarders of the emergency message and it will contend to make the rebroadcasting in case that no vehicle made the forwarding.
In some cases the last segment may have insufficient number of vehicles i.e., the number of the potential forwarders must have a threshold; the number could be generated and tested by:
SucPer is the success percentage that the last segment must fulfill before agree on the values of MaxB and MinB, NeiN is the total number of neighbor vehicles in NT.
If SucPer>Nmax% this means that the last segment holds a sufficient number of potential forwarder vehicles, we subtracted 1 vehicle from the result ofas the last segment also holds the preselected potential forwarder.
If the SucPer<Nmax% this means that the area of the last segment should be expanded to include more number of vehicles, this could be done by:
This will double the size of the last segment, and will increase the possibility of increasing the number of the potential forwarders, if the SucPer still < Nmax% we can calculate MinB again by multiplying Dif by 3 and so on, we should know that when the sender decides to broadcasts an emergency message, it should examine the number of neighbors, if the number is more than one neighbor then this protocol could be implemented, if the number of the neighbors is zero, then the sender will broadcast the message without specifying any forwarder, if the number of the neighbors is equal to one then the sender will broadcast the message and specify the forwarder without adding any detail about the boundaries.
Each vehicle inside the last segment computes its contention time and this will be done by the following Eq:
Tc is the contention time that the segment vehicle has to wait before checking the system to see if the emergency message has been rebroadcasted by other vehicle or not, Tslot: Is the system time slot.
Vehicles having largest progress from the sender will have the shortest contention time to wait before testing the channel to make the forwarding, each vehicle will test its progress from the sender by dividing its current position on the maximum distance computed by the sender, the result of this equation will give waiting time for the contending vehicles inside last segment, giving the opportunity for all the vehicles inside the last segment to recover the failure of the chosen forwarder will increase the percentage of sending the emergency message enabling it to reach longer distances.
Enabling just the last segment to contend will eliminate the hidden node problem
as all the potential forwarders has a high probability to sense the rebroadcasted
message in case any forwarder resent the message as all the potential forwarders
located in a small and limited area.
||Sender decides to send an emergency message
||Sender analyzes the danger and decides its code
||Sender creates NT for its neighbors
||Sender decides the next forwarder depending on the furthest vehicle
||Sender computes the last none empty segment boundaries
||Sender creates the message and inserts the values derived from steps 4
and 5 in the message
||Sender broadcasts the message to the network
Receiving message: Vehicles in VANET receives messages all the time,
these messages could be beacon, emergency or service messages. To know the importance
of the message received, each message should be analyzed, if the message is
holding safety critical information then the message code should be 1 or 2,
in this case this message will be given a higher priority to be processed, the
receiver vehicle also has to make sure that this message is not received before
to eliminate the duplication.
Receiver checks the forwarder id if the current receiver id is the same id of the forwarder, the receiver must start preparing itself to do the forwarding.
If the receiver is not a forwarder, then it must compute the distance between its position and the senders position, to see if the receivers located inside the last none empty segment, if it is located in the last none empty segment it will start contending preparing itself to do the forwarding.
||Receiver receives a message
||Receiver checks the code if it is 1 or 2
||Receiver checks if it is not received before
||Receiver checks if its id equals to the forwarder id
||Receiver rebroadcasts the message immediately
||If not a forwarder, Receiver computes the distance between its current
position with the sender and see if the distance falls within the minimum
and maximum boundary
||Start preparing for forwarding
Forwarding steps: The forwarding job is done by a limited number of
vehicles, we cant assign this job to all the receiver vehicles to avoid
the broadcast storm problem (Tseng et al., 1999)
and hidden node problem, the first forwarder should be the first candidate selected
by the sender.
The forwarding steps for it will be as follows:
||Forwarder waits for a random back off time depending on its
Contention Window (CW), the Back off Time used to avoid channel congestion
||Forwarder Senses the channel and sees if this message has been transmitted
from others or not
||Forwarder reserves the channel
||Forwarder broadcasts the message
The contention window for the first candidate forwarder is 15 μs.
Each vehicle inside the last none empty segment is required to contend before
trying to send the emergency message and this could be done by the following
||Vehicle computes CW depending on
||Vehicle contends for a random Back off time depending on its Contention
||Vehicle Senses the channel and see if this message has been transmitted
from others or not
||Vehicle reserves the channel
||Vehicle broadcasts the message
Simulation results: In order to test correctness of our protocol we
made the simulation using the commercial program Matlab®, we
have selected the use of Matlab as the version R2010b provides a complete and
almost real environment for VANET, http://www.mathworks.com/products/vehicle-network/)
that can simulate the VANET channel by using Kvaser and Vector Drivers (http://www.vector.com/vi_canoe_en,,223.html?markierung=canoe),
we have created the messages and signals using CANoe Tool (http://www.vector.com/)
that is dedicated to manage the DBC (Database) files, we also used AWGN channel
to add noise to the signal, the distribution used is Nakagami distribution.
Parameters used in present simulation are summarized in Table 3; all the simulations in this study will adopt these parameters.
In present simulation we have neglected the fair power sharing for beacons factor as we assumed that the power is constant for all the sent messages.
We made our simulation for 10 sec including 200 vehicles in 2 km road consisting of 3 lanes.
The result could be seen in Fig. 3, 4 and
First we have simulated our protocol CBB and tested it in terms of probability
of emergency message reception and we compared its performance against EMDV
(Moreno, 2007) protocol.
|| Simulation configuration parameters
||Probability of message reception of emergency message with
respect to the distance to the sender
The results show that both of the protocols can increase the distance and performance
of the emergency message, but we have found that CBB can achieve better performance
after the first 1000 meter distance which represents the DSRC communication
range for the sender, in the last meters of the communication range we can see
that the signal is getting to be very weak, but when the forwarder takes the
message the signal again getting to be stronger and enabled it to reach higher
distances, the difference between CBB and EMDV that after several tries CBB
never fails to rebroadcast the emergency message, but EMDV failed in some cases.
We also tested the CBB protocol in terms of the channel collision, we saw that
the channel still almost the same during the system performing the simulation,
but after a period CBB causes slightly more congestion to the channel as it
guarantees to distribute and increase the message reception more than EMDV for
longer distances (Fig. 4).
We simulated the delay for sending the whole journey for the message, i.e., the delay for broadcasting the original message and rebroadcasting, the delay results show that the CBB after several broadcasting attempts and during the time has a slightly higher delay as the EMDV just have a delay on the original message broadcasting, but the delay computed for CBB is the delay for broadcast and rebroadcast the emergency message, thats why the delay for CBB is higher, the delay computed in μs.
The delay computed in this simulation is accumulated for more than one message.
||Collision measured after sending the emergency message during
||Delay measured after sending the emergency message during
CONCLUSION AND FUTURE WORK
We presented in this study an emergency message broadcasting protocol called Contention Based Broadcasting (CBB) protocol which enables the emergency message to be propagated for higher distances to warn the away vehicles about the danger, in the simulation we compared the CBB with the EMDV and the results show that CBB makes better performance for longer distances in terms of emergency message reception, we also found CBB scores a little bit higher collision as the message reception for it is higher during the time, the simulation also showed that for some times the EMDV is low in delay as it is not making the forwarding all the time. In our future work we will test and enable automatic adjustment for the signal power and see its effect on the channel performance.
This paper is financially supported by Institute of Postgraduate Studies (IPS) Fellowship Scheme, Universiti Sains Malaysia (USM). We also would like to thank Mr. Mostafa Taha from Electrical Engineering Dept. Assiut University, Egypt for his valuable assist in the simulation.