1 Introduction

Mobile ad hoc networks offer unique benefits and versatility in wireless environments as well as its applications [14]. It does not require any fixed infrastructure, which includes base stations and prerequisites as well [5]. But, at the same time, it is more vulnerable than the traditional wired network due to its dynamic nature and inadequate protection system [6, 7]. The vulnerability has increased, when the technological advancements has undergone the transition from a single-hop network to a multi-hop network [8]. It has also been considered as one of the primary challenges due to the network’s inherent design [9]. It is intrinsically fault-resilient as the boundary of operation is not limited with respect to topology, delay constraints [1012] and the internet usage [1317]. It is very challenging to secure ad hoc routing because of the difficulty in maintaining any centralized policy or scheme of the traditional network [18]. Many adhoc routing protocols have been proposed in the past [1923]. They have supported dynamic infrastructure that are unpredictable and instantly changing [24]. However, very few proposals have targeted the security requirements. [25] Most of the proposals have been found to inherently trust all the participants in the network, thus making it highly vulnerable [26, 27].

In contrast, the security threats can be from multiple sources such as, external attackers [14], intruders and greedy nodes of the network [28]. The design of routing protocol was earlier supported using an ideal loss-free channel model. Yet, the external factors such as, environmental noise, fading and collisions often degrades the quality of the links [2931]. Besides the above- mentioned difficulties, the resources in MANET also serve as the major constraints. The reason is that they create challenges, while deploying the security processes. They have become sensitive, since the state-of-the-art routing mechanisms emphasize green communications and energy efficiency [32]. Such protocols have insisted spatial re-usability and hence, achieve high end-to-end throughput [33]. AODV [34] and DSR [35] are the two widely known reactive routing protocols that are very efficient in routing, but both are subjected to variety of attacks.

In this paper, we present a Secure Way Routing (SWR) protocol through modifying the conventional AODV routing protocol to address the security challenges in MANET. In SWR, each route to the destination is secure with a unique session key. Using unique session key for each route is a novel contribution of this work. It provides a security model, in which the messages involved in communication are secured using symmetric cryptography. Additionally, the way routing is authenticated with asymmetric cryptography. It secures data routing using a unique secret key that is generated using the session key of the route. This will provide a clear performance advantage such as, high throughput, low end-to-end delay and limited routing overhead. To evaluate the proposal, we have compared SWR with the other protocols like, AODV, Authenticated Routing for Ad hoc Networks (ARAN) [36], Secure AODV(S-AODV) [37] and Stable Route AODV (SR-AODV) [38].

The rest of the paper is organized as sections. Section 2 describes about the related works, which provide an overview of few routing protocols such as, AODV, ARAN, SAODV and SR-AODV. In Sect. 3, we explain the secure way routing protocol mechanism in detail. Section 4 gives the discussion on the experimental results and the corresponding evaluations. Section 5 concludes the paper.

2 Related works

A number of adhoc routing protocols have been proposed in [3942], which have security vulnerabilities due to the wide as well as the open communication environment. These vulnerabilities are now common in mobile adhoc routing protocols also. This paper investigates about the ways to overcome these vulnerability issues, while exploring the existing securing protocols and schemes of AODV, ARAN, S-AODV and SR-AODV.

2.1 Adhoc on-demand distance vector (AODV) routing protocol [34]

AODV is a reactive routing protocol for mobile adhoc network, which constructs the route on demand. It offers low network overhead and uses sequence numbers to ensure prevention from routing loop. Basically, it uses three types of messages to perform communication and maintenance and they are: RREQ, RREP and RRER. It uses table driven routing mechanism for routing the data packets to the destination nodes. Securing the routing message is the main concern in AODV routing. It requires an authentication for securing the messages of the sender as well as the receiver. During route request broadcast, each node checks the originator’s sequence number in the RREQ message against the stored information in the routing table. If a node finds a new request, it updates its routing table. For route reply, it checks the destination node’s sequence number, instead of checking the originator’s sequence number and keeps the routing information updated. Any vulnerability attacks will result in routing loops. Besides routing message modification, spoofing and many other attacks are also the serious issues that are relative to the AODV protocol [34].

2.2 Authenticated routing for adhoc networks (ARAN)

Sazgiri et al. [43] have proposed ARAN for securing the routing mechanism from the unauthorized participation, route modification, spoofing, message modification etc. ARAN is based on an on-demand routing protocol, which extends the features of the AODV protocol. It provides route message integrity and non-repudiation as the minimal part of the security policy in MANET.

ARAN proposes security process in three stages, namely, preliminary certificate process, end-to-end authentication and secure optimal shortest path. It uses trusted certificate server TC and public key cryptography to implement the three stages. Each node must acquire certificate form TC before joining the network. The authentication scheme of the ARAN provides protection against route or message modification, fabrication and impersonation. A launch of denial-of-service attack using a group of malicious nodes through simply broadcasting a larger number of route denial packets exhausts the computational resource to verify the signature and to generate the new ones. This drawback of ARAN utilizes extra bandwidth for transmitting the certificate and creates more routing overhead. ARAN also fail to detect the internal attacks as all the nodes in the network trust each other and cooperate to provide a stable communication [44, 45]. Hence, in case of malicious node presence, it might create huge disturbances.

2.3 Secure AODV (SAODV)

Zapata et al. [37] have proposed Secure–AODV to secure the AODV routing protocol due to numerous security vulnerabilities in the protocol. The reason is that it allows a malicious intermediate node for spoofing its identity illegally, modifying the hop count on route request messages and also to fabricate the route error messages. SAODV is an extension of the AODV protocol, which is based on the public key cryptography to provide routing security. It uses RREQ, RREP and RERR as the routing messages that are digitally signed, in order to secure and guarantee the integrity and authenticity. Every time, a node that generates a routing message signs in with its private key and the nodes that receive this message will verify the signature using the sender’s public key to authenticate. The hop count cannot be signed with the sender because it must be incremented at every hop. Therefore, in order to protect it, a mechanism based on hash chains is used. It generates bigger messages due to heavy weight symmetric cryptography that is used for digital signature. Every time, the messages received in the intermediate nodes must verify the signature for authentication. It increases the burden, when the double signature mechanism is used for generating and verifying a single message.

2.4 Secure routing with the AODV (SR-AODV) protocol

A. Pirzada et al. [38] have proposed Secure Routing with AODV (SRAODV) protocol for securing the routing. It works based on the mechanism of key exchange and data protection. It suggests node to node symmetric encryption for all the information in RREQ, RREP and RERR. It uses a group session key mechanism to negotiate with the neighborhood nodes. This protocol design requires each node to maintain additional information about the associated group members and the session key. This makes it less efficient, when the number of nodes in the network increases and it may also interrupt the normal routing for compromising the modification in hop count or destination sequence.

3 Secure way routing (SWR) protocol

In adhoc routing protocol, the nodes exchange information to their neighborhood and constructs a virtual network for routing the data packets to their desired destination. Such information can be easily targeted by any malicious adversary, who intentionally wants to disrupt the functionality of the network. The attackers generally inject erroneous routing information externally to repeat the previous routing messages or to modify the valid routing information and eventually, bring the network down. Sometimes, due to internal attacks, severe damages are produced as these nodes are not up to their initial commitments. Such nodes also can send erroneous information to modify the local view of the network. Usually, it is very difficult to identify the internal attacker because they already have some sort of credentials that everybody believes.

SWR targets both the external as well as the internal attacks that exist in the network due to malicious nodes. It identifies these attacks based on the three security mechanisms, namely, Certificate Acquisition, Secure Route Discovery and Secure Data Routing. It uses Certificate Authority (CA) certificate to identify the internal attackers and uses both symmetric as well as asymmetric cryptography for getting secured from the external attackers. To prevent the routing information from being forged or tampered, we use CA certificate for encrypting the messages.

3.1 Acquisition of certificate

Establishing security association between the mobile nodes is the most difficult part in the ad hoc network. The difficulty is due to the nature of the mobile ad hoc networks, where the predefined architecture for the security one cannot be used. Most works that are related to security association and key distributions have not been addressed well in most of the previous secure routing protocols. One simplest solution is described in [46] for the existence of security association between the source and the destination nodes. A group key exchange is described in [47], which is based on a strong sharing key. But, this approach required static group nodes and in dynamic networks, where the nodes join and leave very frequently, the group key should be updated using a process for all the nodes.

In [48, 49], another security association process among the nodes has been described. Here, any node in the network can issue certificate for the new nodes and uses asymmetric cryptography. This is a strong approach because it does not have any single point failure in the network. But, it can still have vulnerability attacks during the authentication of a new node and issue a certificate as risky, if malicious nodes are already present in the network. In SWR protocol, in order to have an initial security association among the nodes, we distribute the certificates. But, these certificates are obtained from a trusted certified authority (CA) and it has to be loaded to each node I prior to joining the network. This will be an offline process, where each node has to provide their identity to CA to obtain their certificate.

In this approach, any node that tries to possess an invalid certificate illegally can be identified and isolated easily. The certificate issued from the CA for a node N will have a CA public key as CA pub_key , node address as N add , public key as N pub_key and private key as N pvt_key . The certificate is represented as:

$$C_{N} = Enc_{{CA_{pkey} }} \left( {N_{add} ,N_{pub\_key} ,N_{pvt\_key} ,CA_{pub\_key} } \right).$$
(1)

We assume that all the valid nodes in the network would obtain this certificate before joining the network. This process of acquiring the certificate provides basic identification to the node and prevents it from internal malicious attacks.

3.2 Secure route discovery mechanism

Our protocol modifies the AODV routing protocol to provide the secure routing mechanism as given in Fig. 1. AODV is a reactive protocol, which accomplishes its communication through the processes like, route discovery, data routing and route maintenance.

Fig. 1
figure 1

Throughput performance of the proposed protocol, a with no malicious nodes and b 40 % malicious nodes

Whenever a source node N wants to communicate with a destination node D in the network, it initiates the route discovery process through sending an RREQ message. To make the discovery process secure, the SWR creates a session key using Diffie-Hellman algorithm as S key and then, creates the encrypted message signature using SHA1 algorithm as S m_sign and the encrypted message cipher using CA pub_key as E msg . Before broadcasting, the message is encrypted again using CA pub_key as shown in equation-2. The idea of encrypting the message twice makes it highly secure from the attackers, who are both internal and external. The broadcast message with timestamp T can be represented as:

$$M_{rreq} = Enc_{{CA_{pub\_key} }} \left[ {S_{m\_sign} ,E_{msg} ,S_{key} ,D_{add} ,T} \right]$$
(2)

Therefore, the SWR protocol is capable of determining the secure route through making a comparison among the security parameters, while performing the route discovery of each individual node. The mechanism involved in the route discovery process is described below in Algorithm 1.

figure cfigure c

3.3 Secure routing mechanism

On successful completion of the secure route discovery, the source node sends the data packets on the optimal route that is stored in the routing table. Generally, the AODV protocol maintains only one route between the source and the destination. In our scheme also, we maintain the same. This is due to the fact that in multi- route discovery, the expenses get increased with the storing of more route information. Before sending the data packet, the source should make the data packets secure.

To do so, the source node generates a unique secret key as SC Key using the destination Session Key, D SKey of DH algorithm that is received during the route discovery process. It encrypts the data packets using SC Key and routes the packets. Using this mechanism, the SWR protocol is capable of securing its data packets during data routing in a feasible route. The mechanism achieved using method 1 and method 2 of secure data routing is described in Algorithm 2.

figure efigure e

3.4 Security investigation

This section investigates the possible attacks [50] in the route discovery process and routing as well as the countermeasures taken in the SWR to secure routing in mobile adhoc network.

3.4.1 Attacks in route discovery process

  • Route message modification the process of route discovery requires the intermediate nodes to cooperate to discover the route that reaches the destination. An attack on the intermediate nodes may lead to route message modification.

To handle this kind of attack, the SWR encrypts the route message symmetrically using the SH1 algorithm and asymmetrically using the node’s public key. It provides a double shielding for the attackers to pass through, while they perform route message modification and this serves as the novel contribution of this work.

  • Route cache poisoning This kind of attack misguides the node to route the data in an incorrect path. The SWR handles this attack using the session key that is created by both the source and the destination. A malicious node’s broadcast in incorrect paths will have no effect on the route cache. At first, each route requests the message that is highly secured and protected using the session key and the node’s public key. Secondly, the unique session key makes the message to be completely different from the regular route message.

  • Not participating in discovery process Not participating in route discovery or dropping a packet is a passive malicious attribute that will not interrupt the discovery process, until there are non-malicious nodes available in the network. To handle this kind of behaviour, the SWR ensures that each participating node must have an identity and a CA certificate.

3.4.2 Attacks in data routing process

  • Data packet modification During data communication, it is always possible that the intermediate node can introduce false route through modifying the data packet information and allows the throughput to be degraded. The SWR handles data packet modification through the encryption of data packets using the unique secret key during routing. Both the source and the destination nodes create the unique secret key for sending their data packets and informing the acknowledgement messages.

  • Data packet dropping Data packet dropping is a common behavior of the malicious nodes, which impact the performance of the network. To handle this kind of attack, the SWR protocol ensures that only a trusted and a CA certified node must participate in the communication process.

4 Experimental evalaution

To evaluate the proposed protocol, we assume that both the internal and the external types of malicious nodes exist. However, we also assume that most of the nodes present in the network are trustable due to the certification acquisition form CA. We use the node’s public key cryptography to protect the network against the external attacks and the symmetric cryptography encryption for data and message protection from the internal attacks.

We experimentally simulate the SWR protocol using the Glomosim Simulator [51] to evaluate the performance. It provides a scalable and a parameter driven environment for the wireless protocol simulation. We compare the performances of SWR with SAODV [52], SRAODV [37] and ARAN [36] for evaluation.

4.1 Simulation setup

To simulate the protocol, we setup the parameters that are described in Table 1.

Table 1 Simulation parameters

The simulation runs on the Random Way-point model with a speed variation of up to 100 m/s. We perform the simulation in two sets. The first set does not have any malicious nodes, while the second set contains 40 % of malicious nodes.

During the route discovery process, all the nodes behave normally as they are certified. During data routing, we configure the simulator to randomly choose 40 % of the nodes as malicious. It was observed that those nodes, which behave abnormally, try to modify the data packets and drops all the data packets that are routed through them.

4.2 Performance analysis

4.2.1 Throughput

Figure 1 shows the throughput performance of the protocol. All the protocols show similar results in the absence of malicious nodes. The SWR shows an improvisation, when compared to the other protocols in the presence of malicious nodes. The improvisation in the throughput is due to the efficient securing of the data packets from attacks. In the absence of malicious nodes, it shows an average performance due to the cryptography overhead. The SWR achieves 25 % improvisation in packet delivery, when compared to the other protocols. The other protocols show a downfall of 10–20 %, when 40 % of malicious nodes are present.

4.2.2 End-to-end delay

Figure 2 shows the end-to-end delay comparison between SWR and other protocols. All the protocols show similar ratio of increase in delays with increase in the mobility speed during the absence of malicious nodes. But, in case of malicious nodes’ presence, the SWR and the ARAN show low delays in comparison to other protocols. Both the ARAN and the SWR follow the process of certificate acquisition, which allows secure and identified node in network to minimize packet drop and end-to-end delay in case of malicious attacks.

Fig. 2
figure 2

End-to-end delay performance of the proposed protocol, a with no malicious nodes and b 40 % malicious nodes

4.2.3 Routing overhead

Figure 3 shows the comparison of routing overhead between the SWR and others protocols. In the absence of malicious nodes, all the protocols have similar ratio of overhead. But, in case of malicious nodes’ presence, the SWR shows low routing overhead in comparison to others. This is because the SWR encrypts and decrypts the data packets only at the source end and the destination the end during data communication. On the other hand, in the other protocols, the security checks are performed during communication and the routing overhead gets increased.

Fig. 3
figure 3

Control overhead performance of the proposed protocol, a with no malicious nodes and b 40 % malicious nodes

4.3 Statistical analysis

A statistical analysis has been performed to demonstrate the reliability of the proposed protocol. With the similar experimental setup, we have executed the protocol 100 times. Since the mobility follows RWP plan, every execution produces different performance metrics. These metrics have been obtained and they are subjected to basic statistical functions such as mean, median, best, worst and standard deviation. However, the experimentation is carried out, when the mobile velocity is set to 100 m/s. The results are tabulated in Tables 2, 3, and 4.

Table 2 Statistical analysis on throughput at node speed 100 m/s
Table 3 Statistical analysis on end-to-end delay at node speed 100 m/s
Table 4 Statistical analysis on control overhead at node speed 100 m/s

Here, each protocol is ranked based on the accomplished statistical metrics. For instance, SWR is ranked one in mean values of Table 2, because its mean throughput is higher than the other protocols. Similarly, for every function a rank is assigned and the average rank is determined at the end of the metrics. The final rank for each protocol is determined based on the average rank. This final rank provides a near substantial performers and non-performers.

Under both no attack and 40 % attack constraints, SWR secures first rank except the throughput measures under no attack condition. Despite AODV secures first rank in this condition, it has secured last rank under 40 % malicious attacks. This raises the question about the robustness of the protocol. However, SWR is proven under insecure environment also.

5 Conclusion and future work

We have presented the secure way routing (SWR) protocol for mobile ad hoc network, which secure the routing mechanism from both the internal and the external attacks. It has authenticated the route discovery messages using the public key cryptography and has secured the data routing packets using the symmetric cryptography that has made use of the unique session key and the secret key. The experimental evaluations of SWR have shown an improvisation in throughput and routing overhead in case of malicious nodes’ presence in the network, when compared with AODV, SAODV, SRAODV and ARAN. It provides a novel contribution through providing a double shielded security to the routing message and the data packets and hence, the attackers find difficulty in intruding

An enhancement to the protocol can be made in the future to evaluate more sensitive parameters of the protocol, which can affect the cryptography process. From the simulation, it was also observed that the effects of mobility have high impact on the performance of mobile adhoc network. So, one can enhance the protocol in future to handle link failure and to repair the process.