| Abstract | Due to the expeditious improvement in applications using Mobile Ad-hoc Network, data
dissemination is one of the provocative research topics in this aspect. VANET or Vehicular
Ad-hoc network is a widespread and applied in various situations. The prominent attribute of
VANET is the communication between vehicles.VANET is similar to MANET(Mobile Adhoc
network) where nodes follow frequently join and leave policy.We have used an infrastructure
less concept in this research.VANET can be used for the traffic information,accident
alerts,route information etc.The data is disseminated to different vehicles through a number
of different replica nodes in between destination and source nodes.Normally, In a clientserver
scenario the single point of failure is a common problem.So,in order to avoid this
problem the Content Delivery Networks are used to to decrease response time and bandwidth
consumption of the Global network.The main motive of Content Delivery Network(CDN) is
to move the server close to the user(i.e:client).
In the existing studies it is only proven that CDN technique provides the better Goodput
at destination nodes compared to typical OLSR. In this study, we propose to compare the
Average PDR,Delay and Node speed of VANET using OLSR and CDN adopted OLSR by
varying the size of packets.We divide the total packets to be transmitted into different batches
(50,100,150,200,250,300) per transmission cycle in a Topology Control Interval.We then try
to transmit the packets from source to CDN in a Unicast mode.From the CDN nodes the data
is broadcasted to the destination nodes.As,we broadcast the data to destination nodes,the
redundant bits acquired helps in maintaining a stable packet delivery.
After performing the simulations,we could conclude that the CDN-OLSR at 200 packets per
batch provides the highest Average PDR but at the expense of resource.The delay was gradually
decreasing as the number of packets per batch were increasing.The average PDR was
deteriorating as the node speed was increasing.The performance metrics obtained through
proposed algorithm were far better than normal OLSR. |