Black Hole Attack Prevention on AODV Routing Protocol using Clustering Approach (CBAODV) in MANET

— Mobile ad hoc network (MANETs) is a type of wireless network and that are organized without any predefined infrastructure and centralized administration such as base station or access points. Generally, MANETs nodes can communicate directly if they are in each other transmission range; else the relay nodes are forward the packets to receivers in a multi-hop fashion. Due to some nature of ad hoc networks characteristics such as open medium, infrastructureless and dynamic topology, providing security is particularly difficult than other networks. Black Hole attack is one of the major attack and this detection and prevention is still considered as a challenging task in ad hoc networks. The objective of this paper is to measure black hole attack impacts on AODV routing protocol through clustering approach in MANETs. The performance results of the proposed approach compared with AODV to prove the better performance in terms of delivery ratio, throughput and control overhead.

Miss Bhandare A.S. et. al. [10] proposed an approach called detection and defense mechanism against Cooperative Black hole attack. This anti-prevention system checks route reply against fake reply. The significant advantage of this method is that decision about unsafe route is taken independently by source and no any additional overhead required.
Nidhi Choudhary et. al. [11] identified the black hole by maintaining each node with a trust value for its neighbor node. If the trust value decreases the nodes are indexed in the blacklist table.
Ali Dorri et. al. [12] checks the Next_Hop_Node and Previous_Hop_Node of the RREP in order to check the malicious nodes in the path. Data Routing Information table is maintained by the source node to identify the malicious nodes from the network.
Ashish Kumar Jain et. al. [13] modified the AODV routing protocol by ignoring the first RREP packet reaching the source node through RREP caching mechanism.
Anand Aware et. al. [14] proposed a solution to identify the malicious node by using hash function and rejects first RREP from its neighbor and will select the second optimal path. Kriti Patidar et. al. [15] proposed specification based intrusion detection technique in which every individual node monitor the routing behavior of their neighbors for detecting the malicious node.
Vishvas Kshirsagar et. al. [16] proposed method finds the un-trusted (packet dropper) node from the network, if any un-trusted node found, the performance of the network can be improved by eliminate that node using Bayes' Theorem and Prior probability. This mathematical model secure routing in an independent environment because of it uses heuristic rather than deterministic model. Gayatri Wahane et. al. [17] detected cooperative Black Hole Attack using Crosschecking with TrueLink (Timing based countermeasure) in AODV. The simulation is conducted to prove the minimum routing overhead, delay and maximum throughput when number of nodes and pause time more.
Dhiraj Nitnaware et. al. [18] proposed DYMO (Dynamic MANET On-Demand Routing protocol) to mitigate the effects of Black hole attack on the performance of DYMO.

III. IMPROVED MODEL (BLACK HOLE ATTACK PREVENTION ON AODV USING CLUSTERING APPROACH)
The process of improved model can be visualized with two objectives. The primary aim is to design a stable and flexible clustering algorithm namely Weight Based Clustering Algorithm (WBCA) to elect the appropriate node as cluster head and to manage the nearby members. The secondary aim is to find the malicious node in the formatted clusters.

A. Cluster Formation
Cluster is a subset of nodes in a network. Clustering is a process of dividing the network into disjoint or overlapping clusters. Weight Based Clustering Algorithm (WBCA) is proposed to selects the appropriate node as cluster head to manage the nearby members and to prevent the flood of unnecessary packets. In this algorithm, the maximum hop distance from the cluster head to its farthest cluster member is two hops. Each non-cluster head node is managed by only one cluster head which is one of its neighbors within the two hops. For weight calculation, the weight function w(p) is defined to calculate the weight of each node p as follows w(p) = x × a(p) + y × b(p) + z × c(p) Where, a(p) number of member nodes in one-hop, b(p) number of member nodes in two-hop and c(p) number of cluster member nodes within two-hops. According that, the values are assigned as x = 3, y = 2 and z = 1. After the weight calculation is done, each node compares the weight with its neighbors within two hops for cluster head election. The largest weight node will declare itself as cluster head. The cluster head send HEAD_ANNOUNCE_MSG to the neighbors and acknowledged with receiving JOIN_HEAD_MSG for joining the cluster.

B. Malicious Nodes Detection
The methodology of malicious nodes detections are done at two levels. These are.  Malicious nodes detection using Cluster Head: After receiving HEAD_ANNOUNCE_MSG from the cluster head, if any member does not acknowledge with JOIN_HEAD_MSG it will treated as a black hole.  Malicious nodes detection using check points: The cluster head does not send the HEAD_ANNOUNCE_MSG to the neighbors in the particular time interval t then treating that head as a malicious node. In this situation the next largest node will act as CH. These two levels are detects all the cooperative malicious nodes that are try to drop the packet in the network.

A. Simulation Setup
To analyze the performance, several simulations are run in NS2 version ns-allinone-2.26 under the Red Hat Linux version 9.0 operating system. The simulation composed with 100 nodes that are randomly placed in 500 m x 500 m transmission range within 1000 m x 1000 m area. Each simulation is carried out in 100 sec of simulation time. 20 simulation runs are conducted for each scenario. Nodes depend on random waypoint model and the traffic type is CBR. Each source sends 5 packets/sec and the packet size is 512 bytes.

B. Results and Discussion
Initially, both AODV and CBAODV routing protocol performance results are compared with no black hole nodes. As a result we observe that, under normal situation both performances are almost same in all situations. 1) Packet Delivery Ratio Vs No.of Black Hole Nodes: Packet Delivery Ratio (PDR) is the ratio of the number of delivered to the receivers and number of packet to be received by the receivers. The PDR of CBAODV protocol is compared with AODV protocol as depicted in Fig.1. When the protocols black hole nodes are increasing the CBAODV achieves better delivery ratio as compared to AODV.

2)
Throughput Vs No.of Black Hole Nodes: Throughput defines successful data transmission performed with in a time period and which is normally represented in bytes or bits per second. Throughput of the network is decreases while introducing a node to be black hole node. Although, Fig.2. shows that CBAODV better performances rather than AODV in presence of black hole nodes.

Algorithm:
Step 1: Deploy mobile nodes in the network Step 2: Format the network into different cluster using WBCA Step 3: Identify the cluster heads Step 4: Assign check points for each cluster Step 5: Introduce cooperative black hole in networks Step 6: Level 1: Detection of black hole nodes using cluster head Nodes acknowledge with JOIN_HEAD_MSG with CH. If yes go to Step 7 Else go to Step 8 Step 7: Level 2: Detection of black hole using check points CH sends the HEAD_ANNOUNCE_MSG to the neighbours in the particular time interval t.
If yes go to Step 10 Else go to Step 9 Step 8: Assign node as malicious node and thus node not take part in communication Step 9: Assign the CH as malicious node and go to Step 3 Step 10: Continue packet forwarding till destination is reached.
Step 11: End of the Algorithm. Routing Overhead describes consumed resources in routing process. The better routing protocol overhead must be downward to imply the improved performance. This metric totally depends upon the random topology of the network. We found in Fig.3. that the routing overhead of CBAODV is slightly more as compared to AODV due to the cluster formation. Although, this overhead is acceptable due to it's providing healthier and more affluent delivery ratio. CONCLUSION AND FUTURE WORK AODV routing protocol is one of the best reactive routing protocol for MANET but it is vulnerable to attacks, the major one is black hole attack. The black hole can drop the packets surreptitiously. The paper presents to prevent the AODV routing protocol black holes through Cluster Based Approach. The basic idea behind this approach is formatting the network nodes into clusters to prevent the impact of Black Hole attack regarding performance improvement of the network. The simulation results reveal that the proposed approach better performances in terms of Packet Delivery Ratio, Throughput and Routing Overhead. As a future work, it may extend to identify the attack positions: Malicious nodes are near sender, near receiver or anywhere within the network.