AIRCC PUBLISHING CORPORATION
Consensus Routing And Environmental Discrete Trust Based Secure AODV in MANETs
Radha Raman Chandan and P.K.Mishra Department of Computer Science & DST-CIMS,
The Mobile Adhoc Network (MANET) is a wireless network model for infrastructure-less communication, and it provides numerous applications in different areas. The MANET is vulnerable to a Black-hole attack, and it affects routing functionality by dropping all the incoming packets purposefully. The Black-hole attackers pretend that it always has the best path to the destination node to mislead the source nodes. Trust is the critical factor for detecting and isolating the Black-hole attackers from the network. However, the harsh channel conditions make it difficult to differentiate the Black-hole routing activities and accurate trust measurement. Hence, incorporating the consensus-based trust evidence collection from the neighbouring nodes improves the accuracy of trust. For improving the accuracy of trust, this work suggests Consensus Routing and Environmental DIscrete Trust (CREDIT) Based Secure AODV. The CREDIT incorporates Discrete and Consensus trust information. The Discrete parameters represent the specific characteristics of the Black-hole attacks, such as routing behaviour, hop count deviation, and sequence number deviation. The direct trust accurately differentiates the Black-hole attackers using Discrete parameters, only when the nodes perform sufficient communication between the nodes. To solve such issues, the CREDIT includes the Consensus-based trust information. However, secure routing against the Black-hole attack is challenging due to incomplete preferences. The in-degree centrality and Importance degree measurement on the collected consensus-based trust from decision- makers solve the incomplete preference issue as well as improves the accuracy of trust. The performance of the proposed scheme is evaluated using Network Simulator-2 (NS2). From the simulation results, it is proved that the detection accuracy and throughput of the proposed CREDIT are substantially high and the proposed CREDIT scheme outperforms the existing work.
Keywords: MANETs, Black-hole attack, Security, Discrete trust, Consensus Trust, and Contextual information.
As the Mobile Adhoc Network (MANET) becomes a critical enabling technology, there is an emerging trend in wireless communication security . The characteristics of unguided medium tend the MANET to be more susceptible to various attacks. Among them, the black hole is the most destructive threat to wireless communication. The black hole attackers modify the functionalities of network layer protocol and advertise themselves as the shortest path to the intended node in MANETs [2, 3]. The primary purpose of black-hole attackers is to proclaim as the nearest route to the intended receiver and to drop all the received data packets. Mostly, the Adhoc On-demand Distance Vector (AODV) routing is a widely used network layer protocol in MANETs[4–7]. The source node starts to broadcast the Route REQuest (RREQ) packet if there is no entry for the current path in the routing table for the intended receiver node. On receiving the RREQ packet, the good intermediate routers ensure either if it is the intended receiver or it has a valid route to the receiver node or not. If a correct path has entered in the routing table, a node sends back a genuine Route REPly (RREP) packet otherwise; it continues the RREQ broadcasting. The Black-hole attackers utilize the weakness of the underlying network layer protocol and gives false routing information to launch the black hole attack. The black hole attack deteriorates the routing performance since the routing protocols have generally designed with the assumption of cooperation among the nodes [9–12, 38].
In conventional, there are several trust-based security schemes against the Black-hole attacks in MANETs . The primary process of these techniques is to observe the routing behaviour of MANET nodes in past communications. The trust of a node represents the faith in the routing behaviour of a node. Most of the trust-based security solutions maintain a counter for each neighbouring node and count the packet transmission failures. The transmission failure count of a node increases, only when a neighbour node refuses the data forwarding through a discovered path. By continuously monitoring the neighbourhood activities, the trust management schemes maintain and update the trust value of nodes in MANETs. It is a straightforward scheme to identify the black-hole attack. However, the trust measurement between the nodes does not always represent the actual relationship. Some of the trust-based mechanisms collect the trust values directly and indirectly. The collection of indirect trust information frequently from neighbouring nodes using its direct communication increases the computational complexity and uncertainty to the trust measurement process. Mostly, the trust measurement schemes assume that all the collected trust evidence from the neighbouring nodes are always trustworthy. How- ever, in many cases, it is not reliable. In order to ensure an efficient, secure routing protocol in MANETs without degrading the routing performance, the proposed scheme presents the context-aware routing protocol[40, 43, 44, 41, 42]. This work is organized into the following sec- tions: Section 2 deals with the related works. Section 3 explains in detail about the overview of the Proposed Methodology. Section 4 deals with the experimental evaluation of the proposed algorithm. While Section 5 provides the conclusion and future work for the proposed scheme.
2. Related Works
The Black-hole attack is a packet dropping attack, and such an attack severely deteriorates the routing performance in MANETs. Several defence mechanisms have been proposed to identify and isolate the black-hole attacks in the network.
The Mitigating Black Hole effects through Detection and Prevention (MBDP-AODV) is sug- gested in . Like , it exploits the dynamic threshold value for the destination sequence number. The source node applies mean and standard deviation estimation for multiple RREP packets. In the malicious environment, it is used as the threshold for the destination sequence number. If a source node receives the RREP more than that of the threshold, it is noted as sus- pected RREP. The source node forwards the suspect packets towards the suspected node. When the intermediate nodes receive the suspect packet, and the hop count is equal to one, then the next-hop node is identified as Black-hole attackers. The major drawback of this protocol is that it cannot mitigate the impact of Black-hole attackers when they act as smart in the network. To avoid the effects of Black-hole attacks in MANETs, the AODV is enhanced using watchdog nodes in . Using the number of sent and forwarded packets, the trust of each neighbouring node is estimated by the watchdogs. They are responsible for sharing the reputation value of all other nodes in the network. It improves the routing performance under a malicious environment. However, it is inadequate for detecting smart attackers in large scale networks.
To identify the smart Black-hole attackers, the Timer Based Baited Technique (TBBT) and Detecting Black-hole attacks on MANET by using Harmony Search Algorithm (DBHSA) are suggested in  and  respectively. Each node performs the Baiting phase with random time before sending the data packets to the destination . Within a random time, the source node initiates the bait RREQ into the network. If there is any black-hole node, it replies with the fake RREP packet to the source node. Using such a fake RREP packet, the security scheme identi- fies the attacker node. By executing the phase of the non-neighbour reply, the attacker nodes are removed from the neighbour list. It improves the routing protocol security in MANETs. However, it increases the communication delay drastically. For reducing the delay of the coop- erative bait detection algorithm, the harmony search algorithm is exploited in . By utilizing the data routing information, the Black-hole attacks are identified using the Hybridization of Particle Swarm Optimization with the Genetic Algorithm (HPSO-GA) routing system in . Using PSO and GA techniques, the HPSO-GA improves the security of MANET communi- cation. When an interference causes between multiple routes, the HPSO-GA results in poor performance. In , the source node waits until it receives RREP packets from multiple nodes, after broadcasting the RREQ packet into the network. A first RREP packet is considered as the response from the Black-hole attacker, and the next RREP is accepted for the data forwarding. However, it is not adequate in all the scenarios. If a source node receives the genuine RREQ, it loses the shortest path, as per the suggested security mechanism. It increases the delay and reduces the communication efficiency. Moreover, it does not always protect the network from the Black-hole attackers.
An agent-based technique is proposed in  using Ant Colony Optimization (ACO) algorithm. It exploits the digital signature scheme, watchdog, and path rater techniques to avoid the impact of Black-hole attackers in MANETs. Without having the digital signature, the Black-hole at- tackers do not involve in the route discovery process. Moreover, by applying ACO over multiple RREP packets, the security scheme in  can successfully identify the attackers effectively. How ever, the malicious scenario increases the routing overhead and deteriorates routing performance. Security scheme in  utilizes the reliability factor in the detection of Black-hole attackers in MANETs. If the reliability factor is in confusion state, the fake RREQ is broadcast into the network for preventing the Black-hole attackers. However, it is not the capability of detecting the smart Black-hole attackers, where only the received data packets are dropped partially, but not entirely. Thus, it is essential to consider the entire characteristics of Black-hole attackers to avoid the impact of those attackers on MANET routing.
Liu et al. introduced a trust model for mobile Adhoc networks. This trust model uses both trust propagation and cryptography. In the proposed trust model, every node is initially assigned a trust value. In this paper, the author aims to develop an important concept for establishing a dynamic and collaborative trust model for mobile Adhoc networks. Furthermore, it could be used to enhance the significant measure of trust in the routing of a message in MANET. The author discussed the concepts in this paper are generic. It does not rely on centralized control, any particular routing protocol or key distribution protocol. The proposed method does not need accurate time synchronization, authentication system or any complex hash chaining tech- niques. The proposed method easily integrates with the current routing protocol of the mobile Adhoc network. However, the proposed model does not restrict or modify the route maintenance behaviour or the route discovery of the underlying protocol.
In ,author recommended an effort return based model. This model is maintaining and estab- lishing a trusted route without any cryptography means in the Adhoc network for DSR protocol. In network, every node calculates a direct trust level for all immediate neighbours. For calcu- lating trust information for concerning nodes, every node uses a reputation exchange protocol to share reputation. HashCash a CPU-cost function is used in this model to control the spread of trust reputations by limiting the generation of extravagant requests.
This paper is based on a fuzzy dynamic programming theory. In which a Nobel trust manage- ment model FTDSR(fuzzy trusted dynamic source routing) is proposed. This protocol is used to discover a trustworthy path and mitigate attacks from malicious nodes. The performance
of FTDSR is compared with DSR and TDSR. Results represent that, there is a remarkable improvement in network throughput, PDR and detection ratio for malicious nodes.
2.1 Problem Statement
Most of the conventional security mechanisms assume that the AODV routing protocol is affected by the single black hole attack. These works are inefficient while facing smart Black-hole attackers in the provision of secure wireless communication[31, 32]. It emphasizes the recent research on MANET security to focus on the entire features of Black-hole attackers and extend the conventional systems to mitigate these attacks in MANET. Moreover, it is essential to focus on secure communication without affecting the performance of the AODV routing protocol. In a wireless communication environment, the possibility of occurring the harsh channel condition such as network collisions is high. This impact on Black-hole attack detection is high because the packet dropping due to network collision makes difficulties in the security system[28, 29]. Hence, observing the packet loss rate with sufficient numbers of interactions is essential. For facing such a problem, the proposed scheme plans to integrate the direct and indirect measurement by considering the number of interactions as a weighting factor. Moreover, the entire features of Black-hole attacks such as hop count deviation, sequence number deviation, and routing behavior are considered. It improves the efficiency of the attack detection system in MANETs.
The main contributions of the proposed Consensus Routing and Environmental DIscrete Trust (CREDIT) Based Secure AODV in MANETs are as follows.
3. Overview of the Proposed Methodology
The MANET routing protocols have no techniques to detect the Black-hole attacks in default. This work extends the conventional trust measurement of MANET routing, which adopts only the routing behaviour observation to cope with the malicious activities, by incorporating the discrete and consensus-based trust measurement for formulating the secure intermediate router selection problem as a secure coalition formation. The extended protocol is named as CREDIT. To detect the Black-hole attacks, the CREDIT includes three components, such as Building Dis- crete Trust, Building Consensus Trust, and Aggregating Discrete and Consensus Trust. Figure 1 represents the block diagram of the proposed methodology.
Fig. 1. Block Diagram of the Proposed Methodology.
Fig. 2. Sample Network Scenario.
From figure-2 The Black-Hole attacker 8 claims that it has the best route to the destination 15, but it does not have any route to the destination. The CREDIT is designed to restrict and detect the Black-Hole attacks in MANET. The trust measurement is built by the discrete trust as well as consensus trust collected from the neighbouring nodes.
Table 1. Path List.
In table 1, the paths available between source (node 6) and destination(node 15) are listed along with its corresponding hop count and destination sequence number. The discrete trust measurement relies on the contextual information of a specific node. The contextual information is routing behaviour, hop count deviation, and sequence number deviation. The routing behaviour is the benefit metric, whereas others are cost metrics. Table 2 represents the trust table of node 6.
Table 2. Trust Table of Node 6 .
The cost metrics denote the specific action of Black-hole attackers, since they exploit the smallest hop count, and the highest sequence number to launch the attack in MANET. The cost metric deviation is measured concerning the difference of hop count of a node from the maximum hop count of a neighbouring node for a particular destination. For sequence number deviation, the difference between the sequence number of a node and the minimum sequence number of a neighbouring node for a particular destination is considered. Table 3 represents the normalization of cost metrics of nodes.
Table 3. Normalization of Cost Metrics.
3.1 Building Discrete Trust Model by using Beneftt and Cost Parameters
The Black-Hole attacker claims that it has the best route to any node in MANET, but it does not have any route to the destination. The CREDIT is designed to restrict and detect the Black-Hole attacks in MANET. The design of fully distributed trust on MANET secure routing mandates that each node has to analyze the neighbour list in different perspectives from routing behaviour to contextual information. The trust measurement is built by the discrete trust as well as consensus trust collected from the neighbouring nodes. The discrete trust measurement relies on the contextual information of a specific node. The contextual information is routing behaviour, hop count deviation, and sequence number deviation. The routing behaviour is the benefit metric, whereas others are cost metrics. The cost metrics denote the specific action of Black-hole attackers, since they exploit the smallest hop count, and the highest sequence number to launch the attack in MANETs. The Cost metrics are measured only when a route reply packet is received through the neighbouring node. Notably, only the Benefit metrics denote the behaviour of a node, but the Cost metrics denote the trustworthiness of a route reply routed through a neighbour node. Thus, the Benefit metric based discrete trust is updated continuously, but the Cost metrics are maintained temporarily.
Normalization and Aggregation of Beneftt and Cost Parameters in Discrete Trust Model:- Every node maintains the benefit and cost parameters in a list of the neighbouring nodes. The Cost metric value is zero until an RREP packet is received through the neighbour- ing node. The proposed CREDIT plans to normalize the parameters into a similar standardized format. In the attack detection process, normalization plays a crucial role. The data normaliza- tion is essential, especially when dealing with the parameters of different units. For example, the hop count deviation is a negative metric, and it uses the difference between the number of intermediate routers on a path. However, routing behaviour is a positive metric. The normal- ization of benefit parameter j of neighbouring nodes NH is done using the following equation. The neighbour list in node i and neighbouring nodes of node i are denoted using the NH and NHi notation respectively. Where Vj represents the routing behaviour of a neighbour node in past transmissions, Min and max Vj represents the minimum and maximum value of Vj among all the neighbour nodes of NHi respectively.
Example :Calculation of Beneftt metrics (Routing Behaviour) from table:-2. BV of node 1= (0.9-0.82)/(0.96-0.82)= 0.57; BV of node 5= (0.85-0.82)/(0.96-0.82)=0.214; BV of node 11= (0.82-0.82)/(0.96-0.82)=0; BV of node 13= (0.96- 0.82)/(0.96-0.82)=1;
The normalization of cost parameters k of neighbouring nodes is done using the equation (2). Where Vk represents the hop count deviation and the sequence number deviation, which are observed from the recently received Route REPly (RREP) packets for a particular destination. During the route discovery process, the nodes that have a route to the particular destination replies to the source node with RREP packets. The RREP packet includes the hop count to reach the destination and sequence number. The cost metric deviation is measured concerning the difference of hop count of a node from the maximum hop count of a neighbouring node for a particular destination. For sequence number deviation, the difference between the sequence number of a node and the minimum sequence number of a neighbouring node for a particular destination is considered. If no neighbour node initiates the sending of RREP messages, there is no value for cost metrics in the neighbour list.
Example : Calculation of Cost Metrics (Hop Count Deviation and Sequence Number Deviation) form table:-3.
For Hop Count Deviation:-CV of node 5= (2-2)/(2-0)= 0; CV of node 13= (2-1)/(2-0)=0.5; CV of node 1= (2-0)/(2- 0)=1; For Sequence Number Deviation:-CV of node 5= (998-998)/(998-0)= 0; CV of node 13= (998-0)/(998-0)=1; CV of node 1= (998-0)/(998-0)=1;
The Discrete Trust metric (DT) is associated with the Benefit and Cost parameters. Based on the CREDIT, each node i estimates the discrete trust value DT on the neighbouring node NHi(denoted as DT(i,NHi)) using the following equation:
Example : By fig:2, The Discrete Trust metric (DT) is measured for the one-hop neighbours, which forward the RREP packet for a destination node 15. Calculation of Discrete Trust metric for node 5,13,1 from above calculated value of respective benefit metric and cost metric. for The DT (i, NHi)areasfollows : −DTofnode5 = 0.214 + 0 + 0/3 = 0.0713; DTofnode13 = 1 + 0.5 + 1 = 0.83; DTofnode1 = 0.57 + 1 + 1 = 0.856;
However, the generic trust model without considering the contextual information is not com- patible with the wireless scenarios. The benefit parameter value measurement is inadequate to conclude the routing behaviour of a node. However, packet loss also happens due to communi- cation through a wireless medium. For the wireless nodes, it is insufficient to measure the direct trust with only the routing behaviour. The consideration of Cost parameter values is a benefit for observing the behaviour of Black-hole attackers exactly, the integration of Benefit and cost parameter values improves the efficiency of discrete trust measurement in MANETs.
3.2 Building Consensus Trust Model
The routing behaviour observation by a node is insufficient to confirm the presence of Black- hole attackers. However, it happens only after completing sufficient direct interactions. For the mobile nodes, it is inadequate to measure direct trust with limited communications. Thus, the trusted neighbouring nodes are considered as decision-makers. Concepts of MANET analysis are explained below in Definitions [35, 25, 26, 30].
Deftnition 1: In proposed CREDIT algorithm, the MANET is denoted as a directed graph G(D, E), with the nearby nodes representing decision-makers D ∈ NH = 1, 2, 3, m, Where
m ≤ NH and edges E represents the wireless relationship between node i and decision-makers
∈ NH. The concepts in the MANET are formally described in the following definition.
Deftnition 2: An adjacent matrix A = (DT (i, NHi))m × m is used to describe G(D, E). Where, (di, dj) = 1 denotes that the di has trust value on a node dj. Otherwise, there is no direct relation between the nodes di and dj.
An adjacent matrix denotes whether the trust relationship between node i and decision- maker NH exists or not. However, trust strength is not measured. To solve this problem, the CREDIT proposes an adjacent weighted matrix. An adjacent weighted matrix is denoted using A = (WDTij)m m,dij [0, 1] denotes the trust strength of node i on node j.
An example scenario for the adjacent weighted matrix, associated with the directed MANET is illustrated in table 4. From table 4, the adjacent matrix is built.
Table 4. A Weight Directed Graph .
Table 5. Sociometric .
Deftnition 3: The sequence of edges(di1 , dj2 ), (di1 , dj4 ),(di1 , dj5 ), (di2 , dj5 ),(di4 , dj5 ), (di2 , dj3 ), (di3 , dj4 )—–(din−1 , dj) in Manet G(D,E) are called as trusted links between the decision makers, the trusted link from decision maker di to decision maker dj is represented as di → dj .
In table4, The first five trusted links are used in estimating the discrete and consensus trust of node one on neighbouring node 5.
The weight is measured regarding importance degree. Due to the limitations of the wireless medium, it is difficult for the decision maker to provide a trust assessment on every neighbour- ing node. It tends to incomplete preferences and inefficient trust measurement in MANETs. To overcome this issue, the CREDIT plans to estimate the importance degree of every consensus trust value, which are provided by the decision makers.
Example : In Fig.3, The source node 6 measures the consensus trust on node 5 using the neighbouring nodes.
Fig. 3. Example Scenario for Consensus Trust Measurement .
For the above example scenario, in fig: 3 there are five expert a12, a32…a52 establish the trust relationship across a group in matrix. For figure the adjacent matrix is created, as follows.
Importance Degree Measurement The CREDIT estimates the incomplete preference val- ues of a decision maker by preferences provided by other decision makers. The trust values of other decision makers reflect the importance of other decision makers’ choices in the incomplete preference values estimation. The Importance Degree (ImD) is measured using the equations (4) and (5).
Deftnition 4:The in-degree centrality C(dk) of a trusted link of decision maker dk is con- sidered as:
Example: Applying the values from matrix of ftgure 3, the importance degree of each neighbour is identifted. The in- degree centrality and Importance Degree are measured by for all the neighbouring nodes using the equations (5)and(6). The in-degree centrality are measured for all the neighbouring nodes as follows:-
where C(d15) is the in-degree centrality of a trusted link d15. C(d) of 1 on 5 = (1/(5-1)) (0.9+0.62+0.95) = 0.615; C(d) of 2 on 5 = (1/(5-1)) (0.85+0.62+0.95) = 0.60; C(d) of 3 on 5 = (1/(5-1)) (0.85+0.9+0.95) = 0.675; C(d) of 4 on
5 = (1/(5-1)) (0.85+0.9+0.62) = 0.5921;
The importance degree are measured as follows:-
ImD 1 = 0.615/ (0.85+0.9+0.62+0.95) = 0.1852; ImD 2 = 0.60/ (0.85+0.9+0.62+0.95) = 0.1807; ImD 3 = 0.675/
(0.85+0.9+0.62+0.95) =0.2033; ImD 4 = 0.5921/ (0.85+0.9+0.62+0.95) = 0.178
The incomplete trust values are estimated by the preference of the most trusted decision-maker by the decision-maker who provides the incomplete trust value. In order to improve the trust consistency degree, it is essential to reduce the distance between the estimated trust value and the decision-makers’ weighted preferences using the importance degree measurement.
3.3 Aggregating Discrete and Consensus Trust using Certainty Measurement
Although, integrating discrete and consensus trust is essential for improving the accuracy of Black-hole attack detection in MANETs, considering the number of communication (Com) is im- portant for overall trust measurement. The CREDIT utilizes the certainty degree 1 (Com)−1, which denotes the confidence level of the discrete trust value, concerning the number of commu- nications between the nodes. The confidence level on discrete trust is high when a high number of exchanges are processed. Notably, not only the malicious behaviour affects communication efficiency, but also the network constraints also affect the successful packet delivery. Thus, a large number of communication is essential to estimate an accurate trust value.
Definition 5: Considering the three nodes x,y,z where the node x and y do not have a number of interactions required for trust estimation. However, some information on whether or not node x can trust node z can still be inferred, based on transitivity. Therefore, it is necessary to design a mechanism to analyze whether an unknown expert can be trusted or not. The Einstein prod- uct is used as the triangular norm. The consensus trust is estimated by considering the general equation given below[27, 34, 39].
Table 6. Trust Propagation via Trust Path .
A path is selected, only when a neighbour node attains high trust value due to the routed le- gitimate RREP packet. A highly trusted path is selected through the high trust neighbouring node. It reduces the impact of Black-hole attacks on routing efficiency, due to the consideration of Benefit and Cost metrics in the trust measurement. Thus, the Discrete and consensus-based trust evaluation in MANETs improves communication security without degrading the routing performance.
Example:Using the equation (7), the overall trust is measured. Considering that the communication between node 6 and 5 is 10. and OT(6,5) =1-(10)-10.85 + 1-(1-(10)-1)0.000093992 = 0.765 + 0.000009399; = 0.765009
Likewise, the OT value on nodes 1 and 13 are measured. Among them, a node 13 attains high trust value. Thus, path 2 in 2is selected for data routing. Moreover, node 5 receives less trust among them. So, the RREP initiator of node 8 is considered as a suspected node. Using the CREDIT methodology, the MANET can deliver the data packets successfully to the destination
4. Performance Analysis
Performance Evaluation of the proposed algorithm (CREDIT) under the black-hole attack is im- plemented using the NS-2.35 network simulator. Simulation has been performed for various cases like Varying the normal number of nodes and Varying the number of attackers . Fur- thermore, The proposed CREDIT algorithm is compared with the existing Enhanced AODV (EAODV). The CREDIT algorithm is a modification of AODV which is a standard and widely used routing protocol for wireless Adhoc networks. The performance evaluation is con- ducted over a randomly distributed mobile nodes, and the number of mobile nodes deployed in the network is set to 100. The nodes move with the maximum speed of 30m/s over an area of 600m x 600m and employing the IEEE 802.11 MAC protocol. The nodes can directly com- municate with each others in the range of 200m. The CREDIT uses Constant Bit Rate (CBR) and User Datagram Protocol (UDP) in the application and the transport layer respectively. Performance Metrics and simulation environment7 has given below.
4.1 Performance Metrics and Simulation Result
– Throughput:It denotes the rate of successful data delivery over the communication link in the network.
– Delay: An average time taken by a packet to reach the destination.
– Detection Accuracy: It is the ratio of the number of accurately identified Black-hole attackers to the total number of Black-hole attackers in the
Table 7. Simulation Parameters .
4.2 Simulation Result
Test 1: Varying Number of Nodes In this test evaluation, simulation has been performed with varying no. of mobile nodes 10 to 100 in MANET. In addition, there is a single black hole node which is active in the network, which is performing a malicious activity like packet dropping. For performance evaluation, there are fixed parameters like PDR, Network Throughput and End to End Delay. Simulation results are illustrated regarding the following parameters are given below.
Fig. 4. Varying No off Nodes.
Test 2: Varying the Number of Malicious Nodes Simulation is performed for varying the attacker’s nodes from 2% to 10% with randomly distributed normal mobile nodes set up to 100 in the network environment along with fixed performance matrix parameters. The conclusion of simulation is showed in the form of graphs.
Fig. 5. Varying No of Malicious Nodes.
Test 3: Comparative performance Analysis of CREDIT with existing EAODV Pro- tocol The proposed CREDIT is compared with the existing Enhanced AODV (EAODV). The total simulation time is the 30s. To compare the performance of CREDIT and EAODV, the number of Black-hole attackers is varied from 2% to 10%. This scenario builds low to the high threat environment.
Fig. 6. Varying No of Nodes.
This work presented a trust-based defence system against the Black-hole attackers in MANET. The proposed CREDIT considered the Discrete and Consensus Trust in identifying and iso- lating the Black-hole attackers from the network. It has demonstrated the accurate detection of Black-hole attackers and efficient packet delivery capability of CREDIT in the presence of 2% to 10% of Black-hole attackers in the network. The usage of Benefit and Cost metrics in discrete trust measurement prevents the inaccurate trust measurement due to dynamic network conditions without increasing the routing overhead. The consideration of the certainty factor
in the integration of discrete and consensus trust values decides the dynamic weight for those trust values and improves the accuracy of trust measurement against the Black-hole attackers. Since there is a possibility of fake trust exchanges, the importance degree is measured for every evidence provider. Thus, the CREDIT can detect the Black-hole attackers, even if fewer interac- tions are performed between the nodes. The evaluation of CREDIT protocol shows the improved detection accuracy of the Black-hole attackers by nearly 70% even under a highly vulnerable MANET environment, compared to the existing EAODV.
6. Limitations and Future Work
This work may extend the conventional trust measurement of MANET routing, which adopts only routing behaviour observation to cope with malicious activity. In addition, performance evaluation of proposed work under Black-hole attack has not been performed for varying the mobility of nodes in terms of speed. Furthermore, various performance metric parameters like routing overhead, Normalized Routing Load (NRL), route discovery latency and malicious dis- covery ratio which can be added for examine the performance of protocol in presence of malicious nodes. This limitation may be considered in future work for extension of protocol for better and efficient results.
This work can also be extended for securing some more reactive routing protocol except AODV. Further more, the proposed algorithm can also focus on providing detection accuracy of mali- cious nodes using a suitable and effective statistical method.