PERFORMANCE ANALYSIS OF RANDOMIZED REVERSE AD HOC ON DEMAND DISTANCE VECTOR ROUTING PROTOCOL IN MANET

In high mobility, routing in Mobile Ad-Hoc Network is a very difficult task. In AODV and RAODV, all the data packets are travel through the same shortest path so the intruders can easily trace out the data path. The main objective of Randomized RAODV provides the multipath and then the paths are selected randomly for security purposes. Using randomized routing algorithm to choose the path randomly and then the data packets are travel through different path to reach the destination, so the hackers cannot know about what are the ways the data packets traverse. The performance of proposed RRAODV is compared with the existing routing protocol like AODV, RAODV in mobile network environment. Performance metrics such as packet delivery ratio, end to end delay and control packet overhead are evaluated using NS-2 based on the number of nodes and speeds. Simulation results shows RRAODV gives better performance than the existing protocols like AODV and RAODV for the above metrics. RRAODV is helpful to increase the performance of data transmission and security of data.


INTRODUCTION
A mobile Ad-hoc Network is a collection of interrelated nodes with no infrastructure and a multi hop wireless network with no centralized administration. The nodes in a mobile Ad-hoc network change dynamically such as some nodes join in a network, disconnect the network and also move at any time on the network. In the above reasons, the routing problem in MANET is more difficult than compared to the wired network.
Routing protocols in Ad-hoc network is grouped under three categories like proactive, reactive and hybrid routing protocols. In proactive type of routing protocols, the routing information in each node of a network is updated periodically whether the routing path is needed or not. DSDV and OLSR are example of proactive routing protocol.
Examples of Reactive routing protocols are AODV, DSR and TORA. This type of routing protocol maintain route only for data communication. These type of protocol helps to reduce routing overhead. Zone based Routing Protocol (ZRP) is an example of hybrid routing protocol. It combines the features of proactive and reactive routing protocols.
The rest of the paper is organized as follows. In the next section describes related work, third section describes proposed work, fourth section explains the Science Publications JCS simulation results and parameters. Finally, conclusion is added in the last section.

RELATED WORK
Split multipath routing (Lee and Gerla, 2001) is used to construct maximally disjoint paths in ad hoc networks. The destination node receives multiple RREQ packets and then selects the two maximum disjoint paths. AOMDV (Marina and Das, 2001) computes multiple loop-free and link-disjoint paths in AODV for route discovery but it cannot maintain the alternate path properly.
NMN-AODV protocol (Zangeneh and Mohammadi, 2011) based on AODV protocol and it improves the packet delivery ratio, reduces end-to-end delay and also uses two node disjoint routes between source and destination pair. MMDV (Mtibaa and Kamoun, 2006) protocol provides multipath and MPR based flooding in AODV and this protocol establishes multiple and disjoint path in a single route discovery process but cannot easy to handle congestion and collisions in high node density.
NDMR (Li and Cuthbert, 2004) protocol to reduce routing overhead and also achieves multiple nodedisjoint routing paths in AODV but it decrease the performance when increase the routing load. MP-AOMDV (Sambasivam et al., 2004) protocol form the multiple path and it validate all the alternate path periodically using periodic update packets. Kuo et al. (2009) author describes dynamic routing in DSDV for security purposes. The path between multiple sources to their multiple destinations is stored on a routing table and the paths are selected randomly by the source node and then the data packets are sent through this path.
RAODV  is the extension of AODV protocol. This protocol is helpful to decrease the loss of RREP packet and also prevent a retransmission of RREQ by the source node when the link is disconnected, so the congestion is reduced in a network. Jaisankar and Saravanan (2010), authors compute multiple routes in a single route discovery and it produce node disjoint path and fail safe path for multiple routes. Path Hopping Based on Reverse AODV for security  is the extension of RAODV. In this protocol, the source node stores multiple paths to the destination node and then the paths are selected sequentially from the list. SecMR (Kotzanikolaou et al., 2005) protocol together with AODV for security purposes and it discovers the complete set of non-cyclic and node-disjoint path. Taleb and Behzad (2012), the author performs the simulation study to compare the number of hops in selected path along route reply of AODV and RAODV. They conclude that RAODV data packets meet fewer hops in chosen path and remind energy is higher than AODV. Khelifa and Maaza (2010), author computes the residual energy of nodes in RAODV, EA-RAODV (Gouda and Behera, 2012) is the extension of RAODV which is based on the combination of least hops, power and minimum remaining energy. RAODV (Das, 2013) has better performance than AODV for larger network size and low density networks with lower network mobility. Zarei et al. (2008) develop an algorithm and it is based on link/route stability estimation for decrease overhead of discovery and maintenance of routing. Humaira et al. (2011), Authors compare the performance of AODV and RAODV using the parameters like Throughput, Delay and packet delivery ratio and conclude that RAODV has better performance than AODV.
An enhanced DOA (Vanitha and Parvathavarthini, 2013) to reduce the problem of position estimation error along with DTS and to estimate the nearest neighbor selection.

PROPOSED WORK
This protocol is the extension of Reverse-AODV protocol and it is based on distributed routing information. This one is helpful to improve the security of data transmission. This protocol has multipath routing path to the destination node. Multipath is more advantage in large networks and also provides load balancing. It is helpful to establish more than one path between source and destination node. If the link is failed in a network then the data packets are correctly and securely reach to the destination node using alternate paths to the destination node.

Route Discovery
Source node broadcast the RREQ packet to their neighborhood nodes within their transmission range. The content of the RREQ packet is as follows.
Broadcast ID is uniquely identifying the Route Request (RREQ) packet. The neighborhood node receives RREQ packet which is a destination node it prepares Reverse Route Request (R-RREQ) packet otherwise it forwards to the next neighborhood nodes and this process continues finally this packet reach to the destination nod. When an intermediate node receives multiple copies of the same RREQ packet then it accepts the first RREQ packet and drops the remaining RREQ packet (Fig. 1).

JCS
The destination node only generates the R-RREQ packet and it makes this packet only when it receives the first RREQ packet from the source node. The format of the R-RREQ packet is as follows (Fig. 2).
Destination node broadcast the R-RREQ packet within the transmission range. When an intermediate node receives R-RREQ packet, if it already have the same R-RREQ packet then it drops it otherwise it forward to the next neighborhood nodes on the network. Each node maintains the routing table. The content of the routing table is as follows.
Source node collects the R-RREQ packet from the various neighborhood nodes. Based on the information of routing Table 1, source node selects the path randomly from the routing table for security purposes. Applying randomized algorithm to choose the path randomly on the routing table. Source node selects the path randomly on the routing table and it is used to forward the data packets.

Route Maintenance
Hello packet is used to detect whether the link is failed or not. When the link is failed in a network, neighborhood node of the failed link sends RERR packet to the source node.   When the source node receives the RERR packet, it immediately removes the failed path in their routing table and it chooses another path on the routing table. If a single path between the source and destination is available then all the data packets are travel through the same path. Source node generates RREQ packet only for no path is available on the routing table. So it helpful to avoid extra overhead generated by a fresh route discovery and to reduce the route error transmission during route break recovery.
In randomized RAODV, source node selects the available path randomly in a routing table for data transfer to the destination node i.e., source node selects the different path in each time. In this study, compare the performance of AODV, RAODV and Randomized RAODV using the parameters like Packet Delivery Ratio, End to End Delay, Throughput, Packet Loss and Control Packet Overhead. The above metrics are helpful to analyze the performance of the Randomized RAODV. Packet delivery ratio value and Throughput value is high means the performance of the network is high; If the value of End to End delay, Control packet overhead and packet loss is low means degrade the better performance of the protocol.

MATERIALS AND METHODS
Simulations are helpful to evaluate the performance of the Randomized RAODV and also compare the performance of AODV, RAODV and Randomized RAODV. The simulation environment for performance analysis is shown in Table 2.
Detailed performance analysis of AODV, RAODV and Randomized RAODV using the parameters like Packet delivery ratio, Average end-to-end delay, Throughput, Packet loss, Control packet overhead. Two different scenarios are used to evaluate the above parameters. One scenario is varying the number of nodes but speed is constant and other is varying speed but node is constant.

Scenario 1-Network with Varying Number of Nodes
In scenario 1, AODV, RAODV and RRAODV are analyzed using the parameters like packet delivery ratio, end to end delay and Control packet overhead using varying the number of nodes (40, 60, 80, 100, 120 and 140) and the speed of the nodes is constant ( 20 m s −1 ).

Packet Delivery Ratio (PDR)
Packet Delivery Ratio = ∑ Number of packet receive/∑ Number of packet send Figure 3 shows the packet delivery ratio and it is calculated as the ratio of packets delivered to the destination node to the total number of packets generated at the source node. When the number of nodes is minimum 40 nodes, PDR of Randomized RAODV has 96% and others have 52 and 68%. When the number of nodes increased to 100 and 140, Packet delivery ratio of Randomized RAODV is decreased to 66 and 68%. When the number of nodes increases, the mobility of nodes is also high. Randomized RAODV has better performance of the metric PDR for the number of nodes varies.

End to End Delay
End to End Delay = ∑ (arrive time-send time)/∑ Number of connections End to end delay means the ratio of inter arrival time between the two packets to the total packets delivery time. In Fig. 4, End to End delay is very low in Randomized AODV compared to AODV and RAODV. End to End delay of Randomized RAODV has around 97.33% than AODV and 95.5% than RAODV.

Control Packet Overhead
In Fig. 5, Control packet overhead is less in Randomized RAODV compare to AODV and RAODV.
It measures the total number of control packets. Randomized RAODV has around 62% over in AODV and 35% over in RAODV.

Scenario 2-Network with Varying Speed of Nodes
In scenario 2, AODV, RAODV and RRAODV are analyzed using the parameters like packet delivery ratio, end to end delay and control packet overhead using varying the speed of nodes (10, 20, 30, 40 and 50 m s −1 ) with the number of nodes is constant (60).  Figure 6 shows the packet delivery ratio Vs Speed of AODV, RAODV and RRAODV. When the speed of the node is increased, packet delivery ratio is high in RRAODV compare to AODV and RAODV. RRAODV has 300% over AODV and 90.6% over RAODV.

End to End Delay
In Fig. 7, End to End delay is very low in Randomized AODV compared to AODV and RAODV.
End to End delay of AODV is nearly 1.31 ms, RAODV has 0.91 ms but RRAODV has 0.03ms. End to End delay of Randomized RAODV has around 97.4% than AODV and 97% than RAODV.

Control Packet Overhead
In Fig. 8, Control packet overhead is much reduced in Randomized RAODV compare to AODV and RAODV. Randomized RAODV has around 74% over in AODV and 52% over in RAODV.

DISCUSSION
Performance of the AODV, RAODV and Randomized RAODV is analyzed based on the number of nodes is represented in Table 3. Based on the result, Randomized RAODV has better performance on the nodes 40, 60 and 80 in the parameters packet delivery ratio, end to end delay and control packet overhead. When number of node is increased, the performance of the Randomized RAODV has changed.
In Table 3, using the parameter packet delivery ratio Randomized RAODV has 172.83% over AODV and 71.17% over RAODV. In End to end delay RRAODV has nearly 96% over AODV and RAODV. In control packet overhead, RRAODV has around 62% over AODV and 35% over RAODV. Table 4 shows the analysis results of the performance of AODV, RAODV and Randomized RAODV based on the speed. Randomized RAODV has achieved high packet delivery ratio i.e., 300% over AODV and 90% over RAODV using various speed of the network and delay is 97% over AODV and RAODV and also control packet overhead is 74% over AODV and 52% over RAODV.

CONCLUSION
In Randomized RAODV, the source node collects multiple paths from the destination node and all the path information is stored on the routing table based on hop count. The source node chose the path randomly in the routing table for the security purposes. Intruders cannot identify which way the data packets are travel to reach the destination node. The simulation work is based on the metrics for packet delivery ratio, throughput, control Science Publications JCS packet overhead, packet loss and end to end delay. Using these parameters, Randomized RAODV has better performance than AODV and RAODV with respect to the number of nodes and speeds. But bandwidth consumption is high in Randomized RAODV due to multiple path selection. Our future work will focus on introducing various attacks and then monitoring the performance of Randomized RAODV.