Keywords

1 Introduction

MANET is a type of Ad hoc Network [1] that is formed by self configurable wireless mobile devices: mobile phones, laptops and other mobile devices. These devices are called as nodes. The main character of a MANET is mobile nodes. Source, destination and intermediate nodes are available in MANET [1]. The limited wireless radio range makes source node to depend on number of intermediate nodes to communicate with destination which is not in its‘ radio range. This nature is called as multi-hop routing. Relatively Low cost of network deployment and opportunistic nature enables MANET into many applications. For instance disaster situations, hostile environments and military operations when infrastructure networks are unavailable [1].

Basically routing protocols are categorized into two: Proactive and Reactive. AODV [2] is an example for reactive routing protocol that is more suitable for MANET [1]. Fundamental characters of MANET are dynamic network topology, open network boundary and infrastructure-less nature. These characters are the reasons for the security threats in a MANET [1]. Therefore pure AODV protocol is insecure for data communications. In order to propose a security mechanism following research questions should be answered. I). Does node mobility in a MANET impact the behavior of a malicious node? II). Is node mobility need to be considered when designing appropriate security countermeasures?

Rest of the paper is organized as follows. Related literature are reviewed in Sect. 2 to identify research gap. State of the art of the node mobility related researches are presented on Sect. 3 to point-out the importance of this study. Section 4 describes about the research methodology. Section 5 expresses the results and discussions in our study. Section 6 explains about the conclusion and proposed future works.

2 Literature Review

Camp, T. et al. [3] presented simulation results by using synthetic mobility models which are used in ad hoc network simulation. The data traffic pattern of the mobility model greatly influences the protocol performances. Moreover they [3] argued that Random Way-point Mobility Model (RWMM) is flexible and it creates a realistic mobility patterns and Random Direction Mobility Model (RDMM) is unrealistic than RWMM. The Boundless Simulation Area Mobility, Gauss–Markov Mobility, Probabilistic Random Walk Mobility Models provide movement patterns that one might expect in the real world. Sharma, G. et al. [4] constructed movement paths using the Voronoi diagram of obstacle vertices that is more realistic movement model through the incorporation of obstacles. Connectivity of the nodes, network density, packet delivery and overhead of the routing protocol are effected because of the mobility model. Bai, F et al. [5] did a study and analyze various mobility models and their effects on MANET protocols. Furthermore they [5] attempted to provide current research status of mobility modeling and analysis. They concluded that a requirement method is needed to select a suitable set of mobility models. Lin, G. et al. [6] proposed a technique that enables accurate derivation of steady state distribution functions for node distance and speed to analyze mobility model based on the renewal theory. This theoretical solution is suitable for speed decay problem of nodes in RWMM. Atsan, E. et al. [7] provided a hybrid classification for mobility models. They used Random Walk Model (RWM) and RWMM during their simulation study. Simulation results are discussed based on the simulation settings and parameters. They simulated based on the varying number of nodes and size of the simulation area. RDMM and RWM showed worst performance. AODV protocol performed well for boundless simulation area mobility model.

Sharma, G. et al. [4] designed a research to answer three research questions based on the different mobility models and delay capacity relationship. Their study results showed that the mobility models are considered in the literature are in some sense extreme: they either exhibit the smallest critical delays or the largest critical delays among the mobility models. They introduced an idea of critical delay that is studied to identify the tolerated value to enhance the network capacity. They showed that the critical delay is inversely proportional to the length of the node traveled path in same direction [4]. Radha, S. et al. [8] discussed the survey of about different mobility models that are independent and dependent node mobility. They stated that exponential model performs well than the uniform models for expanded speeds and traffic loads. They stated that best performance is possible with the combination of different mobility models. Ting, Z. et al. [9] studied about node mobility model in vehicular communication system. They suggested to use RWMM and RDMM in a simple environment because of its less computational overhead compared to Vehicular mobility model.

Ciullo, D. et al. [10] analyzed the correlated node movement based on the real mobility process. They enforced the Group Mobility model for better network performances than independent node movements. Through the study, they revealed the wide range of correlated node movements help to better performance than the independent nodes movements. Khairnar, V. et al. [11] evaluated the RWMM through a simulation study in a VANET. They stated that MOVE allows to customize the RWMM in a realistic manner. Vasanthi, V. et al. [12] analyzed some mobility models. Their simulation study is based on the RWMM in DSR routing protocol. The study aims to prove the affect of mobility model on the network performances of the simulation. Finally they concluded that simulation results highly dependent on the node mobility model used in the simulation. Zarifneshat, M. et al. [13] attempted to propose modification on the random mobility models and Levy walk mobility model to enhance the simulation node mobility into realistic ones. Their simulation study is based on random mobility model. Through the simulations results they exposed the correlation between occurrence of an accelerated movement and event of mobile node’s direction change. Han, G. et al. [14] tried to measure the influences of mobility of the selected nodes in an Mobile Wireless Sensor Network on the DV-hop localization algorithm. RWMM, RDMM and Reference Point Group Mobility models are used for the simulation study [14]. They conclude that RD and RWP mobility models perform well in DV-hop localization algorithm. Moreover, mobility models greatly affect the DV-hop localization algorithm. Othman, N. et al. [15] investigate about the node mobility on node cooperation in MANET. In their study they observed that RWMM and Steady-state RWMM impact similarly on the cooperative nodes. Furthermore Self Similar Least Action Walk Mobility model significantly differ from other two models. Finally they concluded that different mobility models produce different impacts on the network.

Though researchers conducted different studies on the node mobility of the network and different mobility models there are no any study considering on the impacts of node mobility of malicious node.

3 State of the Art

As discussed under literature review section, in 2002, Camp, T. et al. [3] surveyed different mobility models for ad hoc network to illustrate the importance of selecting a mobility model for a simulation. Furthermore relationship between mobility model and the network performances is exposed. In 2003, Jardosh, A. et al. [16] proposed a realistic mobility model by introducing obstacles on the route of node movement. Bai, F. et al. [5] surveyed and examined different mobility models that has unique characters such as temporal dependency, spatial dependency or geographic restriction. In 2004, Lin, G. et al. [6] proposed a new methodology to simulate mobility of nodes with a guaranteed steady state for node movement distribution from the beginning. In 2006, Atsan, E. et al. [7] surveyed and proposed a hybrid classification on existing mobility models. In 2006, Radaha, S. et al. [8] presented a survey which helps researchers to select a research model for the simulation study based on the performance evaluation. Furthermore, they concluded that combination of different mobility model may be more effective for simulations. Sharma, G. et al. [4] analyzed different mobility models by using a framework which evaluates a mobility model by critical delay. In 2009, Ting, Z. et al., [9] presented a comparative study of mobility models on VANET. In 2011, Ciullo, D. et al. [10] concluded from their research study that correlated nodes movements lead to better performance than independent node movement mobility models. In 2012, Vasanthi, V. et al. [12] analyzed different mobility models including entity models. In 2013, Zarifneshat, M. et al. [13] mobility trace produced by a mobility trace generator tool. They used association rule mining concept to find possible correlation between mobile nodes. In 2014, Othman, N. et al. [15] investigated the impacts of few mobility models. In the same year, Han, G. et al. [14] quantified the influence of mobility of a node that moved based on three different mobility models by using DV hop based localization model.

4 Methodology

Computer simulation is considered as the research instrument to collect empirical data such as field experiment do [16]. Table 1 shows parameters that are maintained during the simulations. Attacks are introduced individually to the network to check the impacts separately. Impacts on the network is expressed through the PDR, EED and Throughput. PDR is a ratio between total number of received packets from destination node and total number of packets sent by the source node [1]. EED is a time that is taken by a packet to reach the destination node [1]. Units are in seconds. Throughput is a ratio between total number of successfully received packets by destination node and total time taken to receive all the packets [1]. Units are in bytes per second (bps). The attacks are introduced to the network by modifying the functions of AODV protocol [1]. Grayhole attacks is configured as to drop all packets after a certain period of time. Transmission range in between wormhole nodes are maintained as twice the range of the normal node during the wormhole attack simulations. We considered following assumptions.

  • Nodes are identical.

  • Configurations are unique for all nodes except wormhole nodes.

  • Malicious nodes show abnormal behavior during the communication.

Table 1. NS2 parameters
Full size table

5 Results and Discussion

Table  2 contains the average values of each PM of the networks which are affected by different security attacks. In this experiment, node mobility is maintained as zero ms\(^{-1}\) in each network. Average values of PM are obtained by averaging each corresponding PM values. Average PDR, EED and Throughput values of the controller network are 50.58%, 0.76 s and 6746.51 bps. According to the results PDR value is decreased on increasing the number of connected nodes in the network. Though, PDR value is increased at last. EED value is increased for increasing the connected nodes amount. Furthermore Throughput value is decreased gradually on increasing number of connected nodes in the network. The reason is, the packets needed to travel through many nodes while increasing the nodes amount.

Due to the effect of Blackhole attack PDR, EED and Throughput of the network are 0.28%, \(\infty \) and 0.33% compared to the controller network. Blackhole node does not allow any data packets but routing packets. PDR and Throughput values are not zero because Blackhole node allows only routing packets. Therefore the Blackhole attack degraded the network performances than all the other network performances.

Wormhole attack enhanced the network performances than other networks including controller network. Because of the higher speed data transferring wormhole tunnel between wormhole nodes. Therefore data packets quickly transfer through the tunnel. Furthermore, wormhole attack is a passive attack which shows no harm or undetectable harm to the network [1]. PDR, EED and Throughput values are 151.60%, 1.31% and 178.41% compared to the controller network.

The network shows intermediate performances in the presence of Grayhole attack. The performances are lower than controller network and higher than the network with Blackhole attack. PDR, EED and Throughput values are 51.30%, 92.67% and 57.23% compared to the controller network.

Table 2. Network performances in zero node mobility
Full size table
Fig. 1.
figure 1

PDR of the network with Blackhole attack vs node speed

Full size image

Figure 1 shows the graph plotted between PDR of different number of connected nodes in the network vs node speed in the presence of Blackhole attack. According to the simulations results, the link between source and Blackhole node is broken and established again at lower number of nodes with higher speed of node mobility. Therefore graph shows slight increase at that point. In the other cases, link breakages did not occur along with the Blackhole node or no any new links are established after a link breakage with Blackhole node.

Fig. 2.
figure 2

PDR of the network with Grayhole attack vs node speed

Full size image
Fig. 3.
figure 3

PDR of the network with Wormhole attack vs node speed

Full size image

Figure 2 shows the graph plotted between PDR of different number of connected nodes in the network vs speed of node mobility in the presence of Grayhole attack. According to the simulations results, when the number of connected nodes and the speed are increased the link between the source and Grayhole node breaks. Then a new route is established between source and destination node. Therefore PDR is increased. When the speed of node mobility is very high, PDR values are decreased due to the breakage of the existing links among the nodes.

Figure 3 is the graph of PDR value of different number of connected nodes in the network vs speed of node mobility in the presence of Wormhole attack. According to the simulations results when the number of connected nodes and speed of nodes are increased PDR value remains same because of the wormhole tunnel. Though when number of node is low and speed of the node is high the link between the source node and the wormhole node breaks. Therefore PDR value is decreased.

Fig. 4.
figure 4

PDR of controller network vs node speed

Full size image

Graph between PDR of different number connected of nodes in the network vs speed of node mobility of the controller network is shown in Fig. 4. According to the Fig. 4, though the number of connected nodes are increased, PDR value is decreased. This is because of the higher speed of node mobility increases the links breakage. Therefore amount of data received by destination is decreased. Even when the number of connected nodes are decreased, destination can receive data by new links which are created by higher speed of node mobility.

Figure 5 shows the graph plotted between EED of different number of connected nodes in the network vs speed of node mobility in the presence of Blackhole attack. According to the simulation results, destination node did not receive any data packets either number of connected nodes in the network is increased or speed of node mobility is increased. Therefore all the graphs are lie on zero.

Fig. 5.
figure 5

EED of the network with Blackhole attack vs node speed

Full size image

Figure 6 shows the graph plotted between EED of different number of connected nodes in the network vs speed of node mobility in the presence of Grayhole attack. According to the results, when the speed of the node mobility and number of connected nodes in the network are increased links are broken including the link between Grayhole node and the source node. Therefore source node is able to communicate with destination with lowest hop count. Therefore EED is decreased when number of connected node and speed of the nodes mobility are increased.

Fig. 6.
figure 6

EED of the network with Grayhole attack vs node speed

Full size image

Figure 7 shows the graph plotted between EED of different number of connected nodes in the network vs speed of node mobility in the presence of Wormhole attack. According to the results, when the speed of nodes mobility and number of connected nodes in the network are increased links are broken including the link between Grayhole node and the source node. Therefore source node is able to communicate with the destination with lowest hop count. Therefore EED is decreased when number of connected nodes and speed of the nodes mobility are increased.

Fig. 7.
figure 7

EED of the network with Wormhole attack vs node speed

Full size image
Fig. 8.
figure 8

EED of controller network vs node speed

Full size image

Figure 8 is the graph between EED of different number of connected nodes in the network vs speed of node mobility of the controller network. According to the results, when the number of connected nodes and the speed of the node mobility are increased the EED value is decreased. This is because with higher number of node and higher node speed existing links are broken and new links are created. Furthermore data packets are reached the destination through newly created shortest route.

Figure 9 shows the graph plotted between Throughput of different number of connected nodes in the network vs speed of node mobility in the presence of Blackhole attack. According to the Fig. 9, when the number of connected nodes and the speed of the node mobility are increased the Throughput value remains same. This is because of Throughput of RREQ packets. Furthermore There was no any link breakage between source node and Blackhole node or no any new links are created due to the node mobility. Graph between Throughput of different number of connected nodes in the network vs speed of node mobility of the network which is affected by Grayhole attack is shown in Fig. 10. According to the Fig. 10, when the number of connected nodes and the speed of the node mobility are increased the Throughput value is decreased. This is because of the link between the source and Grayhole node is broken and new links are not formed between source and destination node. Furthermore in lower number of connected nodes in the network with higher speed of node mobility new links are established between source and the destination node. Therefore Throughput value is increased.

Fig. 9.
figure 9

Throughput of the network with Blackhole attack vs node speed

Full size image
Fig. 10.
figure 10

Throughput of the network with Grayhole attack vs node speed

Full size image

Figure 11 shows the graph plotted between Throughput of different number of connected nodes in the network vs speed of node mobility in the presence of Wormhole attack. Though number of connected nodes in the network is increased and speed of the node mobility is increased Throughput remains same. The reason is Wormhole node are able to maintain the link with source node. Furthermore the link between source node and Wormhole node is broken with the lower number of connected nodes in the network and with the higher speed of node mobility. Therefore Throughput value is decreased. Figure 12 shows the graph plotted between Throughput of different number of connected nodes in the network vs speed of node mobility of the controller network. According to the simulation results, when the number of connected node in the network and the speed of node mobility are increased, Throughput value is decreased. The reason is the link breakage between nodes. Though in the lower number of connected nodes in the network and higher speed of node mobility Throughput value is increased. Because source node is able to create new and shortest path between destination node.

Fig. 11.
figure 11

Throughput of the network with Wormhole attack vs node speed

Full size image
Fig. 12.
figure 12

Throughput of Controller network vs node speed

Full size image

6 Conclusion and Future Works

Node mobility in the network degrades the MANET performances that differ from one another based on the node speed and direction of the movement. Node mobility leads to occasional link breakage between nodes in the MANET. Therefore link breakage affects the activity of a malicious node. All the nodes including malicious node need to under go for route establishment between source and destination nodes after a link breakage. Therefore node mobility need to be considered when designing appropriate security countermeasures. Because if a routing protocol is equipped with malicious screening mechanisms on route discovery stage then malicious nodes can be screened. Link breakage is favorable for security. Because, It is possible to identify Black hole nodes from the initial screening mechanisms on the routing process. Furthermore, it is helpful to find new shortest path to destination node after the nodes movements. During the simulation we observed that initial route links are maintained in AODV until it breaks. Though, destination node is very close to source node. Therefore, this is opened to another research regarding the existing route verification for shortest path for communication during the communication.