Beruflich Dokumente
Kultur Dokumente
37
Then, the batch of packets will be repeatedly trans- in the network may be increased.
mitted within each TC (Topology Control) interval [3],
where TC interval is the duration time for updating net- 3. METHODOLOGY
work topology by MPR nodes. We analyse our proposed
algorithm using mathematical model, and provide the 3.1 Data Dissemination on VANET using CDN
result in general closed-form solution. Technique
We evaluate our algorithm by simulation using NS3. First of all, for the sake of the readers, the algorithm
The performance evaluation metric used here is success- of the proposed data dissemination on VANET using
ful transmission rate, which is defined as the ratio be- CDN technique [4] is explained briefly here. The detail
tween the number of useful received packets without explanation as well as the pseudo code can be found in
redundant to the total number of packets sent. We [4].
compare our results with that of OLSR. Based on the The proposed algorithm was developed based on OLSR
simulation results, it is obvious that our proposed algo- routing protocol [8] [6] , adopting the pure flooding tech-
rithm provides the better successful transmission rate nique via Multi-Point Relay (MPRs) for reducing re-
than typical OLSR. dundant broadcasting packets. MPR nodes are selected
The organization of this paper is as follows: Section based on original MPR selection heuristics; MPR nodes
II reviews the related work. Section III describes our of any node under consideration will be selected from
methodology. Section IV shows the simulation results one-hop neighbor of that node which covers two-hop
and discussion. Section V is the conclusion. neighbor nodes. That is, any one-hop neighbor node,
which is selected by the other nodes (called MPR se-
2. RELATED WORK lector), is declared as the MPR node. By this selection
Data dissemination technologies were divided into three method, the MPR node,selected by the highest num-
broad categories: geocast/broadcast, multicast, and uni- ber of MPR selector, has the highest reachability to the
cast methods [1]. Most of the approaches for data dis- other nodes.
semination in VANET [1] [9] [10] mainly adopted either In this work, we consider the data dissemination from
one of these methods. In [7], it is shown that the broad- a single source node to multiple destination nodes via
casting scheme is more appropriate to apply in VANET a number of CDN nodes. The algorithm to select the
than the unicast scheme. CDN nodes is described as follows:
In [9], it is shown that to apply broadcasting tech- The MPR selection process is carried out in the same
nique for data dissemination in VANET is to focus on way as MPR selection in OLSR protocol. Here,the MPR
problem of data retransmission. The ACK (Acknowl- node, selected by the destination nodes, which has the
edge) messages are used via beacons to reduce redun- largest coverage area will be declared as CDN node. The
dant messages. broadcasting technique is applied in CDN nodes for for-
The research of broadcast storm, a famous problem warding the received packets. The destination nodes are
happening frequently in urban area and extremely dense categorized based on the distance from source node, into
in rush hour situation, were carried out in [10]. The data 2 types; one-hop and multi-hop away from source node.
pouring (DP) and buffering paradigm were proposed to All destination nodes can receive the packets directly
solve the problems of data dissemination incomplete- from their CDN nodes. Besides that they can receive
ness. These techniques helped all vehicles to receive forwarded packets from their nearby CDN nodes.
complete data, even they just arrived.
M. Guo, M. H. Ammar and E. W. Zegura [2] pre- 3.2 Analysis Model
sented the architecture to support video streaming ap- In this part, we create the analytical model to trans-
plications on VANET. They used a store-carry-and-forward fer data in the city mobility scenario. The structure of
scheme to distribute video contents for the receiver node data transmission in our proposed algorithm is shown
by creating a data forwarding zone. They also proposed in figure 1. We transmit data from a single source node
the forwarder selection algorithm by separating data ve- to multi destination nodes via a number of CDN nodes.
hicle based on the knowledge of vehicle. We transfer the whole data using a large number of
A new protocol focusing on the broadcasting of emer- small-sized packets which are repeatedly transmitted.
gency messages for safety application was proposed in Since, our work is developed based on OLSR routing
[5]. The ACK messages were use for reducing the re- protocol, therefore, the set of CDN nodes will be newly
broadcasting packets. identified in every TC interval. At each TC interval,
In summary, several approaches have been proposed source node selects the CDN nodes from the set of MPR
to improve data dissemination on VANET [1] [9] [10]. nodes. Then, it sends a number of packets to all these
In most of the literature [10], the algorithm to select the CDN nodes. When source node already finishes send-
forwarder nodes by generating new messages to discover ing all packets, it will repeat transmission of the same
the proper nodes were proposed. The control overhead packets to these selected CDN nodes. Each period of
38
Source node
[ ] [ ] [ ]
S
n1 n1 n1 n2 n2 n2 nT nT nT
n1
st th
1 TC interval, r transmission cycle
[ , ]
st st
1 TC interval, 1 transmission cycle
[ , ] Figure 2: Node Behaviour when forwarding
packets to each destination node
Figure 1: A structure of data transmission in j th CDN node (j ) at the rth transmission cycle
city mobility model within the ith TC interval
[i,r]
pj Dm is the probability of successful transmis-
sending all packets is called one transmission cycle. In sion when forwarding packets from the j th CDN
order to analyse the data dissemination technique, it is node (j ) to the mth destination node (Dm ) at the
necessary to define some terms and notations as follows: rth transmission cycle within the ith TC interval
Notation : Here, it is natural that the probability of successful
B is total number of data packets transmission of any path equal the product of each link
T Ci is the ith TC interval (i = 1, 2, ... , T) in the path. Therefore, it is obvious that.
ni is number of packets sent during one transmis- [i,r]
h
[i,r]
ini h
[i,r]
inj
sion cycle within the ith TC interval (i = 1, 2, ... p(xj,m ) = pSj pj Dm (1)
, T)
where
S is source node
Ni is number of CDN nodes within the ith TC nj is a number of successful receive packets at ith
interval (1 6 Ni < Total number of nodes) CDN nodes
[i,r]
j is the j th CDN node (j = 1, 2, ... , Ni ) nj = (pSj )ni at the ith TC interval (j = 1, 2,
Dm is the mth Destination node (m = 1, 2, ... , ... , Ni )
M)
Figure 2 illustrates the node behaviour when forward-
PT is the total probability of successful transmis-
ing packets to each destination nodes. Figure 3 shows an
sion
example of the probability of successful transmission at
p[i] is the probability of successful transmission at
D1 when the number of CDN nodes equal N1 . We can
the ith TC interval
calculate the probability of successful transmission from
p[i,r] is the probability of successful transmission
source (S) to destination D1 at the 1st transmission cy-
at the rth transmission cycle (r = 1, 2, ... , R)
cle within the 1st TC interval by using the equation 1
within the ith TC interval
[i,r] that could be shown as follow.
xj,m is an event of successful transmission from
source (S) to destination node (Dm ) via CDN node
N1
(j ) at the rth transmission cycle within the ith TC [1,1]
[ [1,1]
interval pSD1 = p( xj,1 ) (2)
[i,r] j=1
p(xj,m ) is the probability of successful transmis-
sion of event xj,m
[i,r] We know that, for any event Ai (i = 1, 2, 3, ... , n)
[i,r]
pSDm is the probability of successful transmis- n n n
[ X X
sion when sending packets from source node (S) to p( Ai ) = p(Ai ) p(Ai Aj )
the mth destination node (Dm ) at the rth trans- i=1 i=1 i,j=1;i6=j
mission cycle within the ith TC interval n
(3)
[i,r]
\
pSj is the probability of successful transmission + p( Ai )
when sending packets from source node (S) to the i=1
39
From the above example, we can calculate the total
probability of successful transmission for the 1st trans-
[ , ]
,
mission cycle within the 1st TC interval as
[ , ] [ , ] " #
M
[1,1] 1 X [1,1]
[ , ]
,
p = pSDm (7)
S [ , ] [ , ]
M m=1
" R #
[1] 1 X [1,r]
Figure 3: Probability of successful transmission p = p (8)
R r=1
at D1 when there are Ni CDN nodes (at the 1st
transmission cycle within the 1st TC interval) where R is the number of transmission cycle.
Finally, we have the probability of successful trans- Simulation Tool NS3 version
mission at the 1st transmission cycle within the 1st TC ns-allinone-3.15
interval to D1 destination node, which forward n1 pack- Routing Protocol OLSR
ets via N1 CDN nodes. Distance (km) 4
Total Number of Nodes 100
Number of Destination Nodes 80
N1 1 1
NX N1
[1,1] [1,1] [1,1] [1,1] Number of Hop between Source 1-5
X X
pSD1 = p(xj,1 ) p(xj,1 ) p(xk,1 )
j=1 j=1 k=1 and Destination Nodes
Node Mobility Model City (One-way) model
1 1 X
NX j N1
[1,1] [1,1] [1,1]
Y
p(xj,1 )p(xk,1 ) + p(xj,1 ) Packet Sizes 1000 bytes
j=1 k=1 j=1 Transmission Rates 250 packets/s
(6) Number of Repeated
where h in1 h inj Transmitting packets 250, 500, 1000
[1,1] [1,1] [1,1]
p(xj,1 ) = pSj pj D1 per 1 TC interval
[1,1]
nj = (pSj )n1 ,(j = 1, 2, ... , N1 ) Node Speed (0,5] m/s
40
4.2 Performance Evaluation Metric
In this work, we evaluate our proposed algorithm by
the successful transmission rate which can be defined as 1
follow:
ND 0.6 1000
Successf ulT ransmissionRate = (10)
NT 500
0.4
250
where OLSR
0.2
ND is the number of received packets at the desti-
0
nation node 1 2 3 4 5 >5
NT is the total number of packets transmitted Distance from Source node (hops)
41
from MPRs in OLSR, since these nodes can cover all [4] N. Khaitiyakun, T. Sanguankotchakorn, and
subscriber nodes optimally. By reducing the number A Tunpan. Data dissemination on manet using
of forwarding nodes can decrease the interference and content delivery network (cdn) technique. In
packet collision problem. Information Networking (ICOIN), 2014
According to the simulation results, it is obvious that International Conference on, pages 502506, Feb
our proposed algorithm can improve the successful trans- 2014.
mission rate comparing to typical OLSR. For example, [5] A Khan, Jae-Choong Nam, and You-Ze Cho.
the successful rate of our algorithm at 2-hop distance Ack-based broadcast protocol for vehicular ad hoc
is approximately 70 % while typical OLSR provides a networks. In Ubiquitous and Future Networks
small (closed to zero) successful transmission rate. (ICUFN), 2013 Fifth International Conference
In this work, we propose and also analyse the mathe- on, pages 338339, July 2013.
matics model for forwarding packets from a single source [6] Ou Liang, Y.Ahmet Sekercioglu, and Nallasamy
to multiple destination nodes via a number of CDN Mani. A survey of multipoint relay based
nodes. The solution is provided in a closed form. The broadcast schemes in wireless ad hoc networks.
numerical analysis of our proposed mathematics model Communications Surveys Tutorials, IEEE,
will be carried out in the future work. 8(4):3046, 2006.
[7] S. Panichpapiboon and W. Pattara-Atikom. A
6. REFERENCES review of information dissemination protocols for
[1] Wai Chen, R.K. Guha, T. Kwon, J. Lee, and I.Y. vehicular ad hoc networks. Communications
Hsu. A survey and challenges in routing and data Surveys Tutorials, IEEE, 14(3):784798, Third
dissemination in vehicular ad-hoc networks. In 2012.
Vehicular Electronics and Safety, 2008. ICVES [8] A. Qayyum, L. Viennot, and A. Laouiti.
2008. IEEE International Conference on, pages Multipoint relaying for flooding broadcast
328333, 2008. messages in mobile wireless networks. In System
[2] Meng Guo, M.H. Ammar, and E.W. Zegura. V3: Sciences, 2002. HICSS. Proceedings of the 35th
a vehicle-to-vehicle live video streaming Annual Hawaii International Conference on,
architecture. In Pervasive Computing and pages 38663875, 2002.
Communications, 2005. PerCom 2005. Third [9] F.J. Ros, P.M. Ruiz, and I. Stojmenovic.
IEEE International Conference on, pages Acknowledgment-based broadcast protocol for
171180, 2005. reliable and efficient data dissemination in
[3] P. Jacquet, P. Muhlethaler, T. Clausen, vehicular ad hoc networks. Mobile Computing,
A. Laouiti, A. Qayyum, and L. Viennot. IEEE Transactions on, 11(1):3346, 2012.
Optimized link state routing protocol for ad hoc [10] Jing Zhao, Y. Zhang, and Guohong Cao. Data
networks. In Multi Topic Conference, 2001. IEEE pouring and buffering on the road: A new data
INMIC 2001. Technology for the 21st Century. dissemination paradigm for vehicular ad hoc
Proceedings. IEEE International, pages 6268, networks. Vehicular Technology, IEEE
2001. Transactions on, 56(6):32663277, 2007.
42