Location Verification Technique for Cluster Based Geographical Routing in MANET

1.1Geographical Routing in MANETS

In geographical routing mobile nodes in a network are able to ascertain their positions thanks to certain positioning systems such as a GPS, and a source can utilize some kind of location service to obtain the corresponding destinations. Intermediate nodes achieve forwarding decisions of packets using the knowledge of the immediate neighbour positions, and the source appends the destination’s position in the packet header. In a default setting, the data packets are forwarded greedily through to the neighbour nodes thereby allowing the greatest geographical progression toward the destination. However, whenever such a neighbour does not exist, forward is used to recover the local gap perimeter. In this case, the packets travel across the planar local topology subgraph of the network by using the right-hand rule and progress until greedy forwarding can be restarted again. Although better than a global topology-based routing, imprecise knowledge of a local geographical topology, as well as destination positions, reduces the geographical routing performance. In current geographical routing protocols, each mobile node of the network periodically broadcasts a positional beacon to achieve a local geographical topology. However, in high dynamical scenarios, such proactive protocols create a lot of control overhead information even when there is no traffic through the system, as well as outdated results of the topology. Furthermore, reliance on one-hop forwarding of topological information in these geographical routing protocols leads to sub-optimal forwarding and blind forwarding in current systems (Xiang et al., 2007). In general, geographical routing utilizes node locations to identify viable paths to the nodes. Furthermore, nodes not only know their own locations but also their one-hop neighbours. Since the system follows geographical routing, the destinations are explicitly described geographically (a region or a location), with each packet holding a small amount (O, 1) of routing information in the systems (Rathidevi and Kumaran, 2015).

1.2Location Verification Technique

Malicious nodes can affect the forwarding decisions and a location verification technique (LVT) is utilized to check the accuracy of given node positions collected in the neighbour information table. This technique avoids possible errors in node positions in MANET and it is paramount to verify the node positions in the neighbour before choosing the nodes for forwarding packets in the network. Accuracies of the locations can be assured by position verification technique that utilizes the positioning systems. This technique finds the appropriate neighbours for forwarding the packets. A major bottleneck is the precision of such selected neighbour nodes and the identification of the proper ones. This is crucial in avoiding malicious node selection by hackers when selecting non-neighbours as true neighbours. Neighbourhood discovery process, also called the verification, ensures that the selected node is indeed in the right neighbourhood. Thus, after identifying the reliable and dependable nodes, verification technique is performed since the selected nodes are considered for validating other nodes (Papadimitratos et al., 2008; Basha and Joshna, 2014). Position verification can still result in false positive and false negative rates since there are different types of attacks that can happen in the network. Due to limited message reception in typical wireless network settings, it is in general not always feasible to validate neighbourhood nodes with the neighbourhood discovery approach. Malicious hackers can manipulate the network nodes to disrupt communication between different nodes. Also, this can lead to the precise determination of true neighbourhood nodes surrounding the source node (Papadimitratos et al., 2008). These misleading nodes, incorrectly identified as true neighbourhood nodes, lead to false position details at similar ranges (Basha and Joshna, 2014).

1.3Previous Work and Proposed Solution

In prior research (Muthusenthil and Murugavalli, 2014), a protocol based on location supported cluster based geographical routing (CBGR) is proposed for intermittently connected MANET. In the previous CBGR protocol, degree difference based node value, node mobility, and residual energy are used to choose the head of the cluster. In this setup, there is a node with global positioning system (GPS) and a node with antenna. Moreover, the cluster consists of at least one G-node that selects the cluster head based on the information of remaining energy, speed and mobility of the nodes. In the ad hoc networks, due to the mobility of nodes the cluster maintenance is used to dynamically reorganize and reconfigure the overall cluster, and a store-carry forward model along with the geographical routing protocol is utilized. At the end, a location verification update technique is used to prevent location errors due to routing. By extending the work in Muthusenthil and Murugavalli (2014) and the CBGR protocol, our goal here is to design a position verification technique for the cluster based geographical routing in order to ensure the correctness of location update information. We therefore propose a location verification technique for cluster based geographical routing (LVT-CBGR) with the following features:

3.1Overview of the Architecture

In this work, we study a design with location verification technique for the cluster based geographical routing (LVT-CBGR) in MANET. In this technique, a direct symmetry test (DST) is used to securely verify the location claims. Then a location based encryption scheme is employed to protect the estimated location against the eavesdropping attacks. Note that our network architecture consists of the cluster head ( CH[TeX:] $ \mathit{CH}$) and the cluster member ( CM[TeX:] $ \mathit{CM}$) as the main entities.

3.2Metrics Utilized for Estimation in the Network

We describe the main metrics utilized for network estimation and also further introduce a new noise threshold criteria that is not considered in the prior works (Muthusenthil and Murugavalli, 2017). This helps us to evaluate the correctness of location information based on the direct symmetry test (DST). Table 1 lists the parameters that are presented here.

Next, we recall the main steps in a general clustering approach in the following steps:

3.3Possible Attacks in a Location Based Routing Protocol

We next describe the possible attacks on location based routing protocol considered in this work. Our experimental simulation results in Section 4.2 provide the testing of results for the attacks described here. Generally, the attackers can be broadly classified as malicious and compromised nodes. A malicious node is not involved in accessing the message content. A compromised node behaves maliciously and can access the message content as it is trusted by other nodes in the network. A compromised node can launch several attacks effectively without being detected even by the strong cryptographic techniques. Both the malicious and compromised nodes can launch various attacks against the location aided routing protocols. Some of the common attacks are discussed in this section. Figure 2 demonstrates the attacks on sensor nodes.

Fig. 2.

Possible attacks on sensor nodes in a MANET. Here we show attacks that can tamper the location information of true nodes thereby altering the actual position of a node.

3.4Verification of Location Claims

In order to verify the location claims in the MANET, here we propose to use the direct symmetry test (DST) to securely verify the location claims of cluster members CM[TeX:] $ \mathit{CM}$. In the DST, each CM[TeX:] $ \mathit{CM}$ of the cluster evaluates the communication neighbours with direct links. This means checking if the time of flight (ToF) distances are consistent with (i) one another, (ii) the location information propagated by the neighbourhood nodes, and (iii) a transmission range U. More specifically,

We utilize the DST to securely verify the location claims of the user and a similar usage was done by Fiore et al. (2013). Let Ni[TeX:] $ {N_{i}}$ and Nj[TeX:] $ {N_{j}}$ be the two neighbour nodes within the cluster respectively, ‖Ni−Nj‖[TeX:] $ \| {N_{i}}-{N_{j}}\| $ be the Euclidean distance between the node locations, α be the tolerance value of the node mobility, β be the ranging error, and λ be the position error. Initially, in the DST based test, CHi[TeX:] $ {\mathit{CH}_{i}}$ performs the following:

Note that U refers to the maximum nominal transmission range, and it is greater than the distance at which two nodes can communicate. If there is a mismatch in the above three verification rules, then the neighbour node is marked as defective.

3.5Location Based Encryption Scheme

Once the location of the node is estimated, it is required to protect the location claim from eavesdropping and verify the consistency between the location Q and location claims QC[TeX:] $ QC$. Let τ02[TeX:] $ {\tau _{0}^{2}}$ be the noise threshold for the given estimated location, then we use the following scheme to check and protect.

The noise threshold τ02[TeX:] $ {\tau _{0}^{2}}$ is estimated in Section 3.2.

4.1Experimental Setup and Performance Metrics

We used the network simulator NS-2 (Fall and Varadhan, 2007) to test the performance of our proposed location verification technique for cluster based geographical routing (LVT-CBGR) and to perform comparisons. Table 2 summarizes the simulation parameters used. Our simulation environment consists of the following key components:

4.2Results and Discussion

We next report the performance of the above metrics based on our simulation results with respect to varying the number of nodes as well as the number of attackers.

Fig. 3.

Node vs Delay (s). Comparison of our LVT-CBGR and SLVP performance in terms of end-to-end delay as the number of nodes increases indicating that the proposed network obtains lowest delay (s).

Figure 4 illustrates the packet delivery ratio when there is an increase in the number of nodes. It is clear that our LVT-CBGR technique ushers in superior delivery ratio as against the SLVP approach. At the outset, when the rate is 20, the packet delivery ratio is 30%[TeX:] $ 30\% $ larger than the SLVP approach. But the packet delivery ratio is reduced as and when the rate is stepped up, which also illustrates that the packet delivery ratio and the rate are inversely proportional to each other. But in all scenarios, our novel LVT-CBGR method yields superlative delivery ratio than the SLVP technique.

Fig. 4.

Node vs Packet Delivery Ratio. Comparison of our LVT-CBGR and SLVP performance in terms of packet delivery ratio as the number of nodes increases shows that our proposed network obtains higher delivery ratios.

Figure 5 effectively displays the packet drop ratio for the LVT-CBGR and the SLVP approach. In the case of the proposed technique, the packet drop is smaller than that of the modern approach by around a high 70%[TeX:] $ 70\% $.

Fig. 5.

Node vs Packet Drop Ratio. Comparison of our LVT-CBGR and SLVP performance in terms of packet drop ratio (bytes) as the number of nodes increases. Our LVT-CBGR obtains a smaller packet drop even when the nodes increase to 100.

Figure 6 illustrates the performance evaluation and comparison of the approach and the SLVP method with respect to the packet delivery ratio. As the number of attackers goes up, the packet delivery ratio is reduced, which shows that the attackers are capable of adversely affecting the efficiency in performance of the method. Still, the packet delivery ratio of the novel approach is larger than the modern approach. It is established that the resilience of our LVT-CBGR method is around 34% greater than the SLVP approach.

Fig. 6.

Attackers vs Packet Delivery Ratio. Comparison of our LVT-CBGR and SLVP performance in terms of packet delivery ratio as the number of attackers increases shows that our approach is more resilient to attacks.

Figure 7 shows the performance evaluation and comparison of the approach and SLVP method with respect to compromised communication. As the number of attackers goes up, the compromised communication is increased, which shows that the attackers are capable of adversely affecting the efficiency in performance of the method. Still, the compromised communication of the approach is smaller than the SLVP approach. On analysis, it is established that the resilience of our LVT-CBGR method is around 28%[TeX:] $ 28\% $ smaller than the SLVP approach.

Fig. 7.

Attackers vs Compromised Communication. Comparison of our LVT-CBGR and SLVP performance in terms of communication fraction compromised as the number of attachers increases. Our LVT-CBGR compromises less communication as the number of attackers increase to 30.