Routing protocols function as the obligatory force in MANETs to transfer data outside the physical wireless ranges of the nodes. In hierarchical cluster based routing; cluster head nodes and gateway nodes alone participate in routing decisions. Those nodes may fail to cooperate during route discovery due to selfish or malicious grounds. Hence, imposing cooperation among nodes in MANET to employ a secure route becomes an extremely significant issue. Cryptographic mechanisms can be used, but it acquires a high computational cost and may not categorize the nodes with malicious intention. Therefore, we proposed a dual cluster head based trust aware mechanism as an alternative to cryptographic technique to protect forwarded packets from malicious nodes. Our proposed protocol TWCBRP classifies the network into one hop overlapping clusters with primary and secondary cluster heads, which are accountable for conducting all the routing activities. It constantly assurances the trustworthiness of cluster heads by replacing primary with secondary cluster head, as soon as the former becomes malicious. Cluster members send routing packets only through trusted cluster heads and gateway nodes thus guaranteeing a secure path. The performance of TWCBRP is evaluated with Network Simulator2 and illustrates better performance in terms of packet delivery ratio, throughput, delay, and control overhead when compared to a distributed weighted cluster based protocol (CBPMD).
MANET is a self-configuring, decentralized type of unmanaged (ie., infrastructure less) wireless network with dynamic topology. It does not rely on fixed routers or access points as in the case of infrastructure wireless networks. Instead, each node performs as a host as well as a router and participates in routing process by forwarding data for other nodes. Nodes in MANET use flooding as the basic mechanism for forwarding data and control packets. So, the data is forwarded through intermediate nodes dynamically based on the network connectivity. There are number of characteristics in
MANET. That is, they are appropriate for disaster
situations like natural or human induced disasters,
military
battle-fields,
and
emergency
medical
situations, group communications, civil and business
operations [21]. The nature of the mobile nodes in
MANET brands them extremely susceptible to a variety
of security threats because they
usually own low
computational
resource
well
as
short
radio
transmission range due to the limited battery power
they carry, and they might be moving constantly [
2021 Copyright for this paper by its authors. Use permitted under Creative route
Therefore, there is an inducement for a node to misbehave in a malicious and selfish manner without cooperating with other nodes.
The
intention
of
malicious node is to attack and damage the network.
Similarly, the intention of selfish node is to save its
power, memory and CPU time [
significant
an
unpredictable
node
can
substantial damage and undesirably affect the quality
and reliability of data [
Trust mechanism cryptographic can be
used technique
as
[
an alternative
to Trust mechanism computes trust value on nodes which helps to detect and isolate
malicious and selfish nodes to provide secure data transmission. issue. wreak
By electing single cluster head, it is very difficult to
address the issue of cluster stability [
Broadcasting is a fundamental operation of MANET. This
could be productive only if all nodes operate in a
trustworthy manner. Therefore, establishing and
quantifying behavior of nodes in the form of trust is
essential for ensuring proper operation of MANET [
Indirect trust): Here, trust relationships on nodes are established based on recommendations alone [14]. (iii) Hybrid schemes: In hybrid schemes, the trust on a node is computed based on direct trust experience and recommendations from other nodes [14].
In recent times, there has been considerable effort on various trust computing techniques with respect to MANET [11]. Buchegger et al [12] proposed CONFIDANT (ie., Cooperation of Node’s Fairness In Dynamic Ad hoc NeTworks) protocol for detecting and isolating misbehavior nodes in MANET. In this method, confirmation from direct experiences and recommendations are collected. That is trust relationships and routing decisions are constructed on experienced, observed and forwarding behavior of other nodes. Dynamic Source Routing (DSR) is taken as a base routing protocol in this scheme.
M. Tamer Refaei et al [13] suggested a reputation established mechanism as a means of building trust among nodes. Here a node autonomously evaluates its neighboring nodes based on completion of the requested services. The neighbors need not be monitored in promiscuous mode as in other reputation based methods. There is no need of replacing of reputation information among nodes, thus implicates less overhead. This scheme provides a distributed reputation evaluation methodology that is implemented autonomously at every node in an ad hoc network with the objective of identifying and isolating selfish neighbor nodes.
Haidar Safa et al [14] presented a cluster-based trust aware routing protocol (CBTRP) and it is a kind of reactive on-demand source routing protocol. To make sure safe routing path, the proposed CBTRP scheme first establishes the origin for a trusted environment by providing a trust based mechanism to differentiate trusted nodes from malicious ones. The trust value is computed based upon the information that one node can gather about the other nodes. Then, it organizes the network into one-hop disjoint clusters, whereby every node elects the most qualified and trustworthy node of its 1-hop neighbors to be its cluster-head. Cluster members in CBTRP forward packets only through the trusted cluster heads. Packets from malicious nodes are not processed and no packets will also be forwarded to them. Paramasivam. B et al [15] proposed a secure as well as a fair cluster head selection protocol for improving security in MANETs. This model integrates security factors into the clustering approach for achieving attacker identification and classification. Byzantine agreement based cooperative technique is used for attacker identification and classification to make the network more attack resistant. The nodes that are totally surrounded by malicious neighbors fine-tune dynamically their belief and disbelief thresholds. Venkanna.U et al [17] proposed a methodology to elect a accommodating node as the cluster head node by using key decision parameters such as trust value, remaining energy level, and time of availability values of nodes. Cluster stabilization is achieved by electing two cluster heads in which a secondary cluster head will take the role of the primary cluster head whenever the primary moves out of the cluster. The first step in this model is to structure the problem as a hierarchy for cluster formation. The second step is to calculate the relative local weights of key decision parameters namely TV, REL, and ToA towards the goal. The third step is to estimate relative local weight of each node in the cluster with respective to each decision factor. The fourth step is to determine the overall weight value of each node in the cluster.
Rahul. A et al [18] proposed a cluster based indirect trust mechanism to evaluate the trustworthiness of cluster heads. This model consists of three phases such as interaction phase, request phase and trust evaluation phase. In interaction phase, the member nodes will generate feedback values in the range from 1 to 10 depending on the number of successful interactions between the cluster members and cluster head. In request phase, if any node wants to access a secure connection with any of the service providers (CH), it requests the trust value of all its neighboring CHs. In trust evaluation phase, all the CHs will collect the recommendation values from its member nodes and aggregate the recommended values and issues the final trust value to the requesting node. The requesting node will establish its connection with the CH which has a highest final trust value. 4. PROPOSED APPROACH
Our proposed protocol, which is named “TWCBRP”, is a trust aware dual cluster head based routing protocol to provide a secure and stable routing in MANET. Two cluster heads namely primary and secondary cluster heads are elected in order to maintain route stability. The primary objective is to isolate malicious and selfish nodes through trust computations of nodes for providing a secure routing. a) Trust Computation of nodes in clusters (i) Direct trust computation: Each node computes the direct trust value by analyzing the behavior of its neighbor nodes. That is, the information on the subject of other nodes can be gathered by analyzing the forwarded, received and overheard packets. In TWCBRP, trust between two entities is represented by a 3-dimensional metric opinion [14] as follows: + + =1
= ( , , ) …….. (1), such that Where, denotes node A’s opinion about node B’s trustworthiness, in which, denotes the belief that A holds for B, denotes the disbelief that A holds for B, and denotes the uncertainty that A holds for B. In our proposed protocol, a node monitors other node’s behavior using watch dog mechanism [19] to collect & record all positive (P) and negative (N) events about their trustworthiness. Therefore, the opinion metrics of can be expressed as a function of P and N as follows: =
+ +2 =
2 + +2 ……(2) ………(4) =
+ +2 ……….(3) Where, each of the belief, disbelief and uncertainty values may range between 0 and 1 inclusively. The direct trust value of node B by node A is computed as follows: =
………(5) Every time the number of positive or negative events changes, the corresponding opinion values will be recalculated using equations 2,3, and 4 respectively. (ii) Indirect trust computation (ie., recommended trust) follows: The indirect trust value (ie., recommended trust value) of node B by all its one-hop neighbors is computed as the number of one-hop neighbor nodes of B, the recommended trust value on B by all its one hop is neighbors ‘Ai’ based on their belief factors. (iii) Final trust computation The FTV of a node be governed by both the direct trust value and the indirect trust value. The α part of DTV and β part of IDTV are used to calculate the FTV of a node B. It is computed as, such that α+β = 1 where ………… (7), < 0.5, α = 0.5 and >=0.5, α =1 and β b) Selection of primary and secondary cluster heads Periodically, each mobile node broadcasts a HELLO packet to other nodes that lies within its transmission range to notify its presence and to discover its neighbors. Initially, before cluster formation, all the nodes
may be in un_decided state. During cluster formation, when all nodes have discovered its neighbors, they exchange their weight values through
changes either as cluster_head or as cluster_member.
which lies within the transmission range of more than
one cluster heads becomes a gateway node. The HELLO packet format is given in
Status of the node (0-undecided state/1-cluster head/ 2-cluster member) Node ID Weight value Cluster Head’s Neighbor Table
Cluster Head’s Cluster Adjacency Table In our proposed TWCBRP protocol, primary and secondary cluster heads are elected by computing the weight values of the nodes. Each node computes its own weight using the following weighting function which is based on [WCA], [SWDCBRP]:
Wt (V) = (W1*LQ + W2*RS + W3*BW + W4*MV) where, LQ is the link quality, RS is the residual energy, BW is the available bandwidth and MV is the mobility of the mobile node. A node with highest weight among the other nodes in its transmission range is elected as a primary CH. Similarly; a node having a second highest weight is elected as a secondary CH. The following ICF algorithm is used for cluster formation in the network.
Algorithm-1: Initial Cluster Formation (ICF) algorithm /*At system initiation, let us assume that, each node A in MANET holds undecided state and opinion values as, =0; = 0; =1; =0; = 0; Each node A
maintains the weights of its one-hop neighbors and assume node A is invoking
the ………………… (8), algorithm*/ Input: Set of nodes in MANET Output: Set of clusters ICF( ) Begin Do {
If (B==A) { }
Find a node B with highest weight in its CH set (ie., say B[i] where i=1 to N) if (PCH does not exist in the cluster) {
elseif (SCH does not exist in the cluster) {
elseif ( ( >0.5) or ( >=0.5) or ( == ) ) //A checks the opinion value of B {
if (( B is a cluster member or undecided) && (PCH does not exist in this cluster)){
B changes its status to PCH and accepts A as its member }
elseif ( (B is a cluster member or undecided) && (SCH does not exist in this cluster) ) {
B changes its status to SCH; A becomes member of this cluster; } elseif (B is a PCH of this cluster) {
A sends induced Join_Cluster message to B;
B sends an Accept_Join message to A; A becomes a member of this cluster; } } elseif ( >=0.5) {
Remove B from CH set and continue the loop. } } while ((PCH Not Exists) or (SCH Not Exists)); End; The initial cluster formation (ICF) algorithm is described as follows. Each node computes its own weight value using eqn-8 and broadcasts to its 1-hop neighbors through hello packet. Similarly each node receives the weights of its one-hop neighbors and inserts them in its neighbor table and forms a CH set. If a node A, has no interactions with its neighbor nodes B, initially its belief ( ), disbelief( ) and undecided opinion values ( ) would be computed as 0, 0, 1 respectively using the eqns-2, 3, and 4. Each node A then finds a node B with highest weight in its CH set and checks its opinion values of it. If its > threshold or its ≥ threshold or its belief value ( ) == disbelief value ( ), then node B will either become primary CH or secondary CH based on the need. If its ( ) ≥ threshold, then node B will be removed from A’s CH set. The following table-2 describes the format of one-hop neighbor table (1NT): Table-2 one-hop neighbor table Node ID
Node Status
Cluster ID (CID)
Direct Trust value (DTV)
InDirect Trust Value (IDTV)
Final Trust Value (FTV)
Entry update time (in sec) Each entry in one-hop neighbor table contains information about a 1-hop neighbor and also used to record the opinion about each 1-hop neighboring node. This table is used for cluster formation and route discovery. The following table-3 describes the format of two-hop neighbor table (2NT): Table-3 Two-hop neighbor table Node ID
Node Status
Next Hop
Node By examining the HELLO packets received from its neighbors, a node gathers information about its 2-hop neighbors (ie., 2-cluster away nodes) and stores them Entry update time ( in sec) in this table. This table is used during route discovery and data forwarding. The following table-4 describes the format of cluster adjacency table (CAT): Table-4 Cluster Adjacency Table Cluster (CID)
ID
Gateway ID (GID)
Entry update time (in sec) CAT table is used to keep information about its adjacent clusters. That is, a node records the ID of each of its adjacent CHs and the corresponding gateway node to reach it. This table is used during route discovery and data forwarding. 4.2 Cluster maintenance phase Since, our MANET is vulnerable to attacks, the elected primary CH and secondary CH would become malicious or selfish and affect the network connectivity. In our TWCBRP protocol, at system initiation, cluster formation is done through Initial Cluster Formation (ICF) algorithm with two trust aware cluster heads namely primary and secondary cluster heads. The secondary cluster head after being elected keeps itself in promiscuous mode and overhear the transactions of PCH node. If forwarding ratio of PCH becomes lesser than dropping ratio, SCH triggers PCH node with a LIFE_DOWN message, to carry out the pending transactions of PCH by invoking CH_Change algorithm. Similarly, if forwarding ratio of SCH becomes lesser than dropping ratio, it sends a LIFE_DOWN message to all its one-hop neighbors and invokes the CH_Change algorithm to elect a new SCH node. The significance of CH_Change algorithm is that, it involves only the set of nodes that are within the cluster for local cluster heads updating and does not involve the entire nodes in the network for re-election process. Therefore, it minimizes updating overhead during topological change. The CH_Change algorithm is given below: Algorithm-2: Cluster Head (CH) change algorithm // Let us consider Node A as SCH and Node B as PCH // Let us assume Node A is invoking the algorithm Input: SCH node Output: Change of cluster head CH_Change ( ) Begin //F-Forwarding ratio and D-Dropping ratio If (node weight value of B < Th) or (F(B) < D(B)) then //here Node B is PCH {
SCH sends a LIFE_DOWN message to PCH to relinquish the role of PCH and to process its
pending transactions and invokes Elect ( ) function to elect a new SCH;
PCH joins the cluster as a cluster member; } else
If (node weight value of A < Th) or (F(A) < D(A)) then //here Node A is SCH
SCH sends a LIFE_DOWN message to all its member nodes and invokes Elect ( ) function; function Elect ( ) {
//here Node B is cluster members table (ie., say B[i] where i=1 to N)
Find a node B with highest weight in its neighbor If ((node B is a cluster member or undecided) &&
Node B changes its status to SCH and sends a >=0.5)) {
neighbors; } elseif ( <0.5) { } while (SCH Not Exists);
Remove B from CH set and continue the loop. } a) Route discovery Route discovery is a mechanism whereby a source node S wishing to send a packet to destination node D, it is done through intermediate nodes. Route discovery in TWCBRP is done through flooding RREQ packets only with cluster heads and gateway nodes. However, in order to isolate malicious nodes from participating in the network, their 1-hop neighbors will ignore all packets received from them, and will attempt to find a route that does not include intermediary misbehaving nodes.
For that, each node will keep itself in promiscuous mode to record the transaction of its next hop node [18]. For every successful and unsuccessful transaction, it updates its direct trust value and final trust value respectively. Intra-cluster routing takes place when source node S and destination node D are located within the same cluster. This can be identified by PCH’s 1-hop neighbor table. Inter-cluster routing takes place when the source node S and the destination node D are not located in the same cluster. Therefore, primary CH needs to involve gateway node for data and control packet transmission. A gateway node is a node that lies within the transmission range of both the clusters, and would become members of both clusters. Therefore, a powerful node should be appointed as a gateway node for maintaining network connectivity. In our proposed TWCBRP protocol, among the nodes that lies in the common region of more than one clusters, a node with highest weight and highest trust value (FTV) is elected as a gateway node in order to improve network connectivity. The elected gateway node will act as a “trust guarantor” for the cluster heads that lies within its transmission range. b) Data forwarding mechanism in TWCBRP protocol When source node S attempts to send a data packet to packet directly. Otherwise, S checks its 2-hop neighbor can be reached through more than one-hop neighbors, on its cluster status and the information available in the RREQ packet header. It is expressed as follows: 1. If IM is a cluster member or with undecided status, it simply drops the RREQ packet. 2. If IM is a cluster gateway (CGW), it checks whether it is listed as an entry in RREQ packet header. If no, it simply drops the packet. If yes, it unicasts the RREQ to the
corresponding neighboring CH as recorded in RREQ. 3. If IM is a CH, it appends its CID in the traversed cluster address list and increases the NUM2 counter by 1. If D is found to be a 2-hop neighbor, IM unicasts the RREQ to D based on its 2-hop neighbor table. 4.
Otherwise, for each neighboring cluster which is not listed in neighboring CH list, IM records the CID of neighboring
CH and the
corresponding gateway address to reach that cluster in RREQ, increment NUM1 counter by 1, and broadcasts to them. 5.
If no such neighboring cluster is found, it drops 6. Before recording any node’s ID in RREQ packet, each node checks that the recorded entry does not have The following table-5 shows the format of RREQ
<0.5. packet. neighboring CHs are used by gateway nodes to forward the RREQ, and each CH appends its addresses in the traversed cluster address list field in the
a) Simulation Model and Parameters
The Network Simulator (NS-2) is used to simulate the proposed architecture. In the simulation, mobile nodes are randomly deployed in 750 meter x 750 meter region for 50 seconds of simulation time. All nodes have the same transmission range of 250 meters. The simulated traffic is Constant Bit Rate (CBR). The simulation settings and parameters are summarized below: Number of Nodes: 100 to 500; Node Speed: 5 m/s to 25 m/s; Area Size: 750 X 750 m; Mac: IEEE 802.11; Transmission Range: 20m; Simulation Time: 50 Sec; Traffic Source: CBR; Number of CBR connections: 10; Packet Size: 512; Rate: 50 kb; Initial Energy: 20 Joules; Transmission Power: 0.660; Receiving Power: 0.395. b) Performance Metrics
The proposed TWCBRP is compared with the CBPMD protocol [16]. The performance is evaluated mainly, according to the following metrics. Packet Delivery Ratio is the ratio between the number of packets received and the number of packets sent. Packet Drop refers the average number of packets dropped during the transmission. Delay is the average end-to-end delay measured in seconds. Energy Consumption is the amount of energy consumed by the nodes to transmit the data packets to the receiver. Throughput is the average number of packets received per second.
A. Based on Nodes: In our first experiment we vary the number of nodes as 100,200,300,400 and 500. c) Results Figure 7 shows the delay of TWCBRP and CBPMD techniques for different nodes scenario. We can see that, for nodes 100, the delay of TWCBRP is 98.32% lower than the existing CBPMD technique, for nodes 200 the delay of TWCBRP is 96.23% lower than the existing CBPMD technique, for nodes 300 the delay of TWCBRP is 92.61% lower than the existing CBPMD technique, for nodes 400 the delay of TWCBRP is 78.77% lower than the existing CBPMD technique, for nodes 500 the delay of TWCBRP is 93.86% lower than the existing CBPMD technique. In over all we can conclude that the delay of our proposed CBPMD approach has 92% of lower than CBPMD approach. Figure 8 shows the delivery ratio of TWCBRP and CBPMD techniques for different nodes scenario. We can see that, for nodes 100, the delivery ratio of TWCBRP is 27.42% higher than the existing CBPMD technique, for nodes 200 the delivery ratio of TWCBRP is 21.57% higher than the existing CBPMD technique, for nodes 300 the delivery ratio of TWCBRP is 43.66% higher than the existing CBPMD technique, for nodes 400 the delivery ratio of TWCBRP is 84.91% higher than the existing CBPMD technique, for nodes 500 the delivery ratio of TWCBRP is 81.43% higher than the existing CBPMD technique. In over all we can conclude that the delivery ratio of CBPMD approach has 52% of higher than CBPMD approach. than the existing CBPMD technique, for nodes 400 the residual energy of TWCBRP is 34.03% higher than the existing CBPMD technique, for nodes 500 the residual energy of TWCBRP is 22.32% higher than the existing CBPMD technique. In over all we can conclude that the residual energy of CBPMD approach has 23% of higher than CBPMD approach. Figure 10 shows the throughput of TWCBRP and CBPMD techniques for different nodes scenario. We can see that, for nodes 100, the throughput of TWCBRP is 39.26% higher than the existing CBPMD technique, for nodes 200 the throughput of TWCBRP is 34.37% higher than the existing CBPMD technique, for nodes 300 the throughput of TWCBRP is 52.86% higher than the existing CBPMD technique, for nodes 400 the throughput of TWCBRP is 87.37% higher than the existing CBPMD technique, for nodes 500 the throughput of TWCBRP is 65.08% higher than the existing CBPMD technique. In over all we can conclude that the throughput of our proposed TWCBRP approach has 56% of higher than CBPMD approach.
In this paper, we have proposed a trust sensitive weighted dual cluster head based routing protocol which ensures secured routing and enhances connectivity in MANET. Since malicious and selfish nodes are isolated from the routing path, this guarantees secured and trusted path from source to destination. Moreover, with primary and secondary cluster heads, the stability of the routing path as well as the stability of the cluster structure is also guaranteed.