An Efficient Cluster-Based Approach to Thwart Wormhole Attack in Adhoc Networks

Mobile Ad-hoc networks is an ascertaining domain with promising advancements, attracting researchers with a scope of enhancements and evolutions. These networks lack a definite structure and are autonomous with dynamic nature. The strength of the Ad-hoc network lies in the routing protocols making it an apt choice for transmission. With several types of routing protocols available our focus is on LGF (Location-based Geo-casting and Forwarding) protocol that falls in Position based category. LGF assures to grab the attention with its feature of low bandwidth consumption and routing overhead at the cost of unvolunteered attacks resulting in compromising the security of data. In our approach, we present a technique to overcome the profound attacks like Wormhole and Blackhole by aggregating LGF with k++ Means Clustering aiming at route optimization and promoting security services. The proposed mechanism is evaluated against QoS factors like End to End delay, Delivery Ratio, Load balancing of LGF using Simulator NS3.2 which envisioned drastic performance acceleration in the aforementioned model. Keywords—MANET; LGF; K++ Means clustering; network security; wormhole attack; blackhole attack; secured algorithm


I. INTRODUCTION
MANETs, a group of nodes that provide communication through wireless links without a predefined infrastructure and exhibiting dynamic nature has been the choice of practitioners and researchers for two long decades. The ad-hoc feature of MANETS makes it a favorable choice in several applications like Vehicular communication handling various disaster scenarios, defense, Security, and Online meetings. These applications depend on information exchange between nodes that play a vital in the process of Communication. The Crucial component acting as the backbone for node elucidation and improving the strength of MANETS ate routing protocols. It is a syntactic rule for defining a methodology to be undertaken by routers for transmission of data.
Based on the consideration surveyed by various researchers these protocols are classified as Topology based and Position-based. The former protocols rely on the respective structure of the network, whereas the latter originate with the location information of nodes. Topology based routing protocols fall into three well-known models like Proactive, Reactive, and Hybrid. Proactive protocols as the name reflect works based on prior information stored in the table in contrast reactive build a route on-demand when a request triggers. Bridging the gaps among both Hybrid protocols intersects the characteristics of Proactive and Reactive [24]. Few well-known protocols falling on the Proactive side are DSDV, FSR, OLSR, and reactive are AODV, DSR, and TORA. ZRP, ZHLS, CEDAR are occupied under hybrid Class [23]. These protocols fail to outperform when the network turns to be densely populated with a huge number of nodes resulting in large network sizes, thereby lowering performance [20]. To leverage and sustain the network efficiency even with dense networks, MANETS impend on position-based routing protocols with urging requirement of security features [15]. Few protocols of interest are namely LAR, LGF, and Landmark. SLAR is also proposed to provide security against different attacks [16]. These protocols ensure efficient performance when clustered into zones [4]. Position based class mainly emphasizes the position of the node in the network and their performance is analyzed based on qualitative characteristics like Loop free, Decentralized operations, Path strategy, Performance metrics, Scalability, Reliable Delivery service, and Robustness. These Protocols support few strategies in packet forwarding namely Greedy Forwarding, Constrained directional flooding, and additionally Hierarchical or multilevel methods [2]. The greedy method of forwarding works by using optimization criteria for selecting the next node for the transmission of messages [22]. With directional flooding sender floods, packets to nodes toward the direction of destination satisfying predefined constraints and the Hierarchical method works when huge network scaling is on-demand [26].
Among several position-based routing protocols, our focus is on LGF that targets the reduction of routing overhead and bandwidth. In LGF the neighboring nodes in the forwarding zone perform rebroadcasting route request packet and acknowledge the source with route reply. Unlike all routing protocols, LGF is also vulnerable to serious attacks like Wormhole and Blackhole attack. These attacks exhibit an adverse impact on the performance of location-based ad-hoc networks [8].
Our paper enlightens all the fore-mentioned issues and proposes an enhanced approach based on the K++ Clustering technique to overcome attacks in LGF.

III. EXISTING APPROACH
LGF: LGF with its beneficiary factors like lowering bandwidth and packet dropping rises to be the best choice for leveraging performance about measuring concerns like efficient packet forwarding in MANET's. Steps involved in LGF include path discovery and message forwarding [13,22].
 The process initiates from a source with a multicast PREQ packet to all neighboring nodes based on the IP address of the destination. The protocol limits its range within a predefined distance.
 RREQ packet is forwarded to all the neighboring nodes with a distance less than the source node to the destination.
 The process repeats until the RREQ packet reaches the destination that further acknowledges the path to the source node from various intermediate nodes.
 Finally, the optimal shortest path is captured, and intended packets are transferred among source and destination.
Despite limitations with LGF when the range increases, it also suffers from a Wormhole attack that targets the shortest path with an illusion perspective. To handle this perturbation our algorithm fusions LGF with a clustering approach resulting in a secure, reliable transmission. www.ijacsa.thesai.org 2) Exposed wormhole attack: Here the attackers introduce themselves into the network with route discovery technique, thereby exposing Wormhole nodes and hiding liable nodes between source and destination [28]. Based on these modes different forms of Wormhole attacks are-encapsulated packetbased, Wormhole attack, out-of-band path, relaying of packets, and Protocol manipulation wormhole attack [21].

C. Black Hole Attack
A serious problem endeavoring wireless sensor networks is the Blackhole attack characterized to absorb everything that comes on the way thereby decreasing the performance of the network [17]. In this attack, a malicious node or attacker node announces that it indexes the shortest path to the destination resulting in packet loss. It succeeds in communication failure among wireless networks and base stations. This attack results in topology modification including packet damage with forged routing information [23,10].

A. Lgf with Clustering Approach
As LGF is prone to several attacks discussed and even restricted with range constraints. We propose a variation of LGF in combination with the clustering technique to handle the demerits of LGF. K++ Clustering is embedded in our proposed mechanism to strengthen the LGF for overriding the deficiencies [11]. K++ Means: This algorithm aims at clustering the given dataset into clusters and mainly focuses on seed or initial value selection, as an input to k-means. It overcomes the poor Clustering results of K-means which is an NP-hard problem. K-means gives the worst results for super polynomials in input and bad approximation of objective function in comparison with optimal clustering [5] that is overridden in k++ by the defined procedure to initial Clusters [1,3].
Step by Step Procedure for k++ is given as:

1) Select a data point C randomly.
2) For every data point P, Calculate the distance d(p) between P and C.
3) Pick a new data point based on weighted probability distribution with n proportional to d(p) 2. 4) Iterate steps 2 and 3 until optimal k centers are selected.

5)
Continue with k-Means Clustering after the initial seed value is selected [27].
Position Based Protocols by virtue depends on the location of the node which ascertains performance assurance. These Protocols rely on three main sources to identify the exact location of the nodes, namely, based on signal strength, coordinates between nodes, and GPS based node location. Our approach uses the coordinates to locate the point of the node which helps in identifying malicious node location that promotes traffic bypassing within the network.

B. Phases in Proposed Approach:
Phase1: The network is divided into clusters using the k++ Clustering technique, resulting in k value. The source node initiates the RREQ packet for route discovery when a communication link is needed. This RREQ packet is forwarded to the nearest neighbors with the shortest Euclidean distance.
Phase 2: In this Phase Destination Node acknowledges the RREQ packet with RREP through the shortest path opted. The legitimacy of the RREP packet is judged or evaluated based on the predefined threshold value of RREP packets permitted.
Phase 3: Here a node is considered malicious once it violates the threshold value constraint of the RREP packet. The path the malicious node resides is excluded from the transmission path for forwarding packets. This phase also checks the hop count in the routing table to identify the path established by the hidden malicious node.
Our procedure succeeds in overcoming the Range constraint using k++ and Node Authentication by considering the location of the node as criteria using coordinates for calculating the Euclidean distance. Euclidean distance also adds to overcome the shortest path illusion injected by Wormhole and Black attacker nodes with the help of the RREP packet threshold value.

C. Algorithm for Clustering enhanced LGF
There are some descriptions as given below to recover a safe RREP Packet through intermediate nodes.

V. PERFORMANCE EVALUATION AND ANALYSIS
To Implement the Proposed approach using NS3.2, the Simulation parameters are initialized as shown in Table I. The performance of LGF with K++ Means is evaluated by considering the parameters like Load Balancing, End to End Delay, and Delivery ratio.
1) End to End delay: End to End delay is defined as the time incurred to travel from source to destination [19].
2) Packet delivery ratio: This is the ratio of the number of packets delivered to the number of packets sent by the source node [14].
3) Load Balancing: A parameter that defines an efficient distribution of transmission load during transmission in a network. Fig. 1 shows a clear comparison of the reduced End to End delay factor proposed in comparison to the existing system.
Here proposed system is indicated as lgfc-delay with a green spike. Fig. 2 shows a comparison of the Delivery ratio which shows the packet loss of the Proposed and Existing system. Here proposed system Promises a reduced packet loss with a green spike.  Fig. 3 shows a comparison of the load balancing factor which projects a better performance by the pfroposed system of lgfc indicated with a green spike.
Hence the simulation results of the proposed protocol outperform in terms of End to end delay, load Balancing, and in the reduction of Packet Loss by providing node authentication, Reliability and stability thereby leveraging the performance of the network in Position-Based Routing Protocols.

VI. CONCLUSION
This paper aimed to discuss the importance of MANETS and concentrated on the adverse effects of Wormhole and Blackhole attacks in the position-based routing protocol.
LGF Protocol is studied for various setbacks related to delivery, avoiding Attacks, and providing Authentication of nodes with Location as a constraint. Our approach is considered a clustering-based method to overcome the Prior mentioned issues in LGF with enhanced K++ Mean's supporting attack free and secure packet transmission in Wireless Ad-hoc Networks.