Prevention of Hello Flood Attack in IoT using combination of Deep Learning with Improved Rider Optimization Algorithm
Introduction
IoT devices are used for collecting the information and identify hazards successive to tragedies and also for localizing the affected people [1]. However, IoT is not holding up the existence of tragedy, which could be a very important thing for transferring catastrophe readiness with each other by active information, for example, the prediction of disaster and also caution to the systems [2], etc.
Flooding attack is a specific kind of DoS attack existing in MANETs, in which the malicious nodes imitate the legitimate nodes in all situations except the paths discovered by them. The discovered paths are usually intending to use the processing resources of the remaining nodes. This kind of attack is very easy to execute using on-demand routing protocols such as AODV [3], [4]. Some of the flooding attacks are HELLO, DATA, and RREQ. The most dangerous flooding attack is RREQ since it is very simple to generate a disaster of RREQ packets and lead huge damage. The HELLO flooding attack is related to the network layer attack [5], which targets the routing protocols. The protocols require the nodes used for communicating the HELLO packets for reporting to the neighbours they are existing. The node is considered as the local node when a node receives the data packet from another node within its range.
Many research studies have been introduced in the field of RREQ flooding attacks, which significantly concentrates on detection models that depend on the frequency of RREQ packets for sending the information [6]. In order to detect an attack, each node makes use of constant or dynamic threshold value. This threshold value is computed on the basis of the count of RREQs derived by the node for each unit time. If the node receives more RREQ, then that neighbour node is denoted a as malicious node. The above-mentioned models have some defects in dealing with the dynamics of networks. If the flooding attack is well known in wired networks [7], [8], [9], the solutions are not directly implemented for securing mobile DTN, which is having transient connectivity and not accessible at any time. Several solutions are developed for preventing the flooding attack existing in DTN. In [10], a lightweight validation scheme was proposed. In [11], the maintenance of dropping strategy for node’s message queue. In [12], [13], the internal flooding attackers were not detected. Other contributions like [14], [15] limit the nodes’ message creation rates and the detection of internal adversaries, which is beyond the rate limits. However, the rate limit method is not reliable for networks because it is not predictable broadcast demands.
The major contributions of the paper are portrayed as follows.
- •
To develop a secure IoT-WSN with several processing steps like Cluster head selection, k-paths generation, HELLO flooding attack detection and prevention, and optimal shortest path selection.
- •
To detect and prevent the HELLO flood attack from the IoT-WSN using the BAU-ROA-DBN with the input feature vectors like Route Discover Time and Inter Route Discovery Time.
- •
To find the shortest route paths in IoT-WSN enabled by the proposed BAU-ROA approach by considering the objective constraints like node trust, distance between the nodes, delay of transmission, and packet loss ratio.
- •
To validate the performance of the proposed and conventional models in terms of convergence analysis, and analysis on latency, normalized energy, and shortest path length.
The entire paper is modelled as shown below: Literature review and the features and challenges of existing flooding attack detection and prevention models are discussed in Section 2. Section 3 specifies the detection and prevention of hello flood attack in medical Internet of Things. Detection and prevention of hello flood attacked in medical IoT are mentioned in Section 4. Moreover, the development of improved rider optimization algorithm is shown in Section 5. Section 6 illustrates the selection of shortest paths by proposed rider optimization algorithms with the computation of energy dissipation. Results and discussions are mentioned in Section 7. The final conclusions of the paper are shown in Section 8.
Section snippets
Related works
In 2019, Luong et al. [16] have suggested a new FADA for MANETs on the basis of the machine learning algorithm. The method was dependent on the data of each node’s route discovery history for capturing the equivalent behaviours and features of nodes that belonged to a similar class for deciding whether the node was malicious. Moreover, FAPRP was suggested by expanding the actual AODV protocol and combining FADA algorithm. The performance analysis of the suggested model was done based on packet
System model
Consider the medical IoT-WSN system with huge numbers of sensor nodes denoted as . The data from each and every sensor node is transmitted to a single sink node or base station represented as . A direct data transmission is carried out between the nodes with wireless connection, in which the communication performs within the radio range. Every node in the network takes of the role of data packet transmission, which will be under the network with dimension and in metres. Clusters of
Creation of k-paths
From the IoT-WSN, the source node and destination node has to be chosen after selecting the cluster head, which is done by a graph structure. By this way, the user can choose the source node and destination node along with ID. Moreover, the shortest path of the network is selected from the generated k-paths. This k-path routing procedure is based on the correlation of finding the minimum count of possible bandwidth or shortest path between two cluster heads in the transmission fields that
Conventional ROA
The inspiration of ROA [22], [23] is based on the cluster of riders that is going in the direction of destination. Consider some of the rider’s group, which are going towards the same target for winning the race. In the conventional ROA, there are four groups of riders such as “bypass riders, followers, overtakers, and attackers”. This algorithm is following some steps that are described below and should be followed by every rider in the group.
Parameter Initialization of rider and group:
Solution encoding
The proposed BAU-ROA is employed for defining the shortest path that chooses the best among the cluster heads from IoT-WSN model after preventing the HELLO flood attack. The solution pattern before encoding is depicted in Fig. 4. Here, the cluster head is indicated by , and the count of clusters is denoted as . The cluster heads that are in progress is considered for choosing the shortest route path after detecting and preventing HELLO flooding attack. Let, is the source node, and
Experimental setup
The developed shortest route path selection with detection and prevention of HELLO flood attack was implemented in MATLAB 2018a, and the analysis was carried out. For performing the experimental analysis, the number of rounds was considered as 800, the number of nodes allotted as cluster heads was 35, and the initial energy of each node was kept constant as 0.02 Joules. The performance analysis of the suggested approach was compared with state-of-the-art WOA [25], DA [26], DHOA [27], ROA [22],
Conclusion
This paper has introduced a novel algorithm for detecting and preventing HELLO flooding attack by optimized deep learning mechanism. This model included cluster head selection; k-paths generation, detection, and prevention of HELLO flooding attack, and optimal shortest path selection phases. After selecting the cluster head, and k-paths, some Route Discovery Frequency Vectors such as Inter Route Discovery Time and Route Discovery Time of every node was defined in order to detect and prevent
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
References (28)
- et al.
Internet of Things (IoT): A vision, architectural elements, and future directions
Future Gener. Comput. Syst.
(2013) - et al.
Detection of flood disaster system based on IoT, big data and convolutional deep neural network
Comput. Commun.
(2020) - et al.
HTTP flood attack detection in application layer using machine learning metrics and bio inspired bat algorithm
Appl. Comput. Inform.
(2019) - et al.
The whale optimization algorithm
Adv. Eng. Softw.
(2016) - et al.
Using dragonfly algorithm for optimization of orthotropic infinite plates with a quasi-triangular cut-out
Eur. J. Mech. A Solids
(2017) - A. Boukerche, R.W.L. Coutinho, Smart disaster detection and response system for smart cities, in: 2018 IEEE Symposium...
- et al.
Real-time flood monitoring and warning system
J. Sci. Technol.
(2011) - et al.
Security upgrade against RREQ flooding attack by using balance index on vehicular ad hoc network
Wirel. Netw.
(2017) - et al.
A novel approach for mitigating route request flooding attack in MANET
Wirel. Netw.
(2018) - et al.
A dynamic threshold based algorithm for improving security and performance of AODV under black-hole attack in MANET
Wirel. Netw.
(2017)