Vampire Attack Mitigation and Network Performance Improvement Using Probabilistic Fuzzy Chain Set with Authentication Routing Protocol and Hybrid Clustering-Based Optimization in Wireless Sensor Network

e most e ective threat for wireless sensor networks (WSN) is Vampire attacks on sensor nodes as they can stretch the network connectivity among them and inuence the network’s energy, which can drain the network. Vampire attack has particular malicious nature of sensor nodes in which they can widely exploit features of combined routing protocol. Fuzzy rules and fuzzy sets are highly optimal techniques in mitigating the vampire attacks of the network, which can quantify the uncertain behaviour of sensor nodes.is study aims to propose a novel technique using a probabilistic fuzzy chain set with authentication-based routing protocol and hybrid clustering technique for data optimization of the network. e suggested approach here employs a fuzzybased chain rule set to combat growing types of vampire assaults using probability formulas. e authentication routing protocol has increased network routing security. e proposed technique (PFCS-ARP_HC) has optimized the energy consumption of network. Simulation for this technique has been carried out using NS2 and experimental results show the performance of the proposed model in terms of throughput of 98%, packet delivery ratio of 89%, energy consumption of 67%, latency of 46% control overhead of 53%, and attack detection ratio of 87.9%.

security vulnerabilities that can jeopardise their correct operation. WSNs with resource-constrained nodes are extremely vulnerable to a number of attacks due to their simplicity. Sensor devices' high resource constraints present significant hurdles to resource-hungry security systems. Because of the hardware limits, security methods must be exceedingly efficient. is is no trivial task [4]. Sensor networks' most valuable resource is energy. In terms of power, communication is very costly [5,6]. e available communication channels for coordinating their attack. As a result, WSN is vulnerable to a number of attacks that could obstruct its operations and negate the benefits of using its services. If a sensor node is compromised, all vital material, data and code stored on that node can be retrieved. An attacker can quickly recover valuable information from packets that are sent. Finally, attacker can send bogus data into network [7,8], possibly impersonating one of sensors, with goal of altering sensor readings or interrupting internal control data (Message Injection). Attackers may collaborate on a system attack by using a few MNs with similar or greater hardware capacity than genuine nodes [9]. ese rogue nodes can be obtained either individually or by capturing and physically overwriting the memory of a few legitimate nodes. In rare cases, participating nodes may have high-quality communications links available to coordinate their attack. As a result, WSN is susceptible to a variety of assaults that might disrupt its operations and nullify the benefits of using its services [10][11][12].
Vampire attacks are tough to identify and avoid because they use protocol-compliant messages. Vampire attacks are not protocol-specific in the sense that they do not rely on individual routing protocols' design or implementation flaws. Effects of a vampire attack: (a) Vampire attacks are protocol-independent. (b) ey do not cause any interruptions in instant availability. (c) Protocol-compliant communications are used by vampires. (d) Send a small amount of data while consuming the most energy. (g) Vampires do not alter or disturb established courses. Vampire assaults can be divided into two categories. ere are two types of attacks: Carousel and Stretch [13][14][15]. e contribution of this study is as follows.
(i) To develop the novel technique in mitigating vampire attacks and improve the network performance using security-based routing protocol and clustering-based data optimization. (ii) Here the data optimization has been carried out using hybrid clustering technique and vampire attack is mitigated using probabilistic fuzzy chain set with authentication-based routing protocol. (iii) Simulation for this technique has been carried out using NS2 and experimental results shows the performance of the suggested model in terms of throughput, packet delivery ratio, energy consumption, control overhead, and attack detection ratio.

Related Works
Numerous works have been carried out in this area and enormous data are available on DoS attacks and their effects on ad hoc networks. At the routing protocol layer, the author [16] investigates resource depletion assaults, often known as "Vampire Attacks," which permanently disable networks by rapidly draining node battery power [17,18]. ey've included information on numerous vampire assaults and how they affect ad hoc networks. ey've also listed a number of ways to protect the network from these threats. Our previous work [19] provides a quick overview of these two procedures as well as the consequences of vampire attacks on them. Some recent study in this topic includes numerous implementations and theoretical work, such as [20], which shows how to tolerate attack by using CH. In the event of a vampire assault, the Cluster Head intervenes and distributes the packet to its target without dropping it. As a result, even in the event of a vampire assault, the message is delivered successfully and reliably. e authors of [21] suggested how to use PLGP, an identifying malicious assault to lessen the vampire attack. e next study [22] looks into the energy issues and assaults that Ad Hoc WSNs and, eventually, future IoT face. It also provides a strategic method to dealing with energy threats that is resilient. e proposed approach in [23] aims to give a method that may be utilized to find vampire attack in WSN. A new trust routing motif is proposed in [24]. Multiagents gather multifactor data and come together to decide on the trust path. e degree of belief in the long-term conduct of alternative entities that is based on the nodes' previous experience is referred to as trust. CAWS and MES-1, a collection of algorithms put together by a researcher [25], were examined. CAWS (cellular automata-based security algorithms) entails key management and secure digital communication under cellular automata rules, requiring minimal memory and simple calculation. e author of [26] first aims to assess these vulnerabilities in terms of router layer battery reduction attacks. Second, it focuses on making changes to current routing protocols in order to prevent packet loss due to vampire attacks during packet forwarding. To protect wireless sensor networks from wormhole attacks, they propose a trust aware distance vector routing protocol (TAODV) in [27]. eir proposed approach was tested utilising experimental results in terms of enhanced PDR, end-to-end delay and node to destination variation. Ariadne, an on-demand routing system, and the LEACH protocol, intended to reduce battery use, are among the approaches and protocols used to resist these assaults. e LEACH methodology involves two phases: steady state and startup. e cluster is established in the steady-state phase based on threshold value, with each node calculating its value by picking a random number between 0 and 1 and broadcasting it to its neighbours. A node with a value less than threshold limit is nominated as CH and rest of nodes join as member nodes. Although the LEACH protocol helps to save energy, it does not ensure secure routing against vampire attacks.
ere are two phases to EWMA (Energy Weighted Monitoring Algorithm) [28]. e cryptographic keys can guard against active attacks from the outside, but they are unreliable in the case of passive attacks that compromise service quality as well as reliability. Optimization techniques based on the behaviour of lions during territorial defence and takeover [29,30], such as the Lion algorithm, were devised to minimise passive attacks and to optimise path selection. e nonlinear system identification solution is implemented. However, solutions to difficult situations are not taken into account. As a consequence, the authors in [31] presented the WOA, which is also an optimization system based on the behaviour of humpback whales. Because the location of the prey is unknown at beginning, the current solution, or prey, is treated as the best response and position is updated in each iteration, despite the fact that network nodes use more energy for calculation.

System Model
is section discusses the proposed design in mitigating vampire attacks of the sensor networks. Here the proposed module has been divided as three parts: first is hybrid clustering in which the nodes has been clustered based on energy level of energy; secondly analysing for vampire attacks by authenticating data transmission path through routing protocol and probabilistic fuzzy chain set with fuzzy rule sets. e overall proposed flow diagram is shown in Figure 1.

Hybrid Clustering of Nodes Based on the Energy Level.
Consider n nodes in a field. Lifetime of node I is denoted by Li. Let the network lifetime, L, be time it takes for first node in network to exhaust its energy. e main goal is to increase L, which necessitates uniformly utilising the energy of all nodes. To avoid drawbacks of both static and dynamic clustering methods, designed hybrid method so that clustering is not done every round. To do so, CHs save their residual energy in their memory at the end of each setup step after clusters have formed. When a CH detects that its residual has fallen below ECH, it sets a specific bit in a data packet ready to be forwarded to BS.

Cluster Setup Phase.
Nodes that will become CH will be chosen during the clustering setup step. CH is chosen based on delay duration of processing. e primary CH will be chosen when the delay time approaches zero. If a node's delay time is assessed to be lower, it has a better chance of becoming a CH and sending a broadcast packet. Packet size, node ID, node location, residual energy, and packet type are all included in the content of the packet. If a node gets packet type 2 from other nodes before delay timer expires, it will become a normal node "N." Additionally, the candidate CH will send messages to sensor nodes through broadcast. Messages from the CH will be verified by the sensor nodes [32][33][34]. If the CH has a limited amount of residual energy and is likely to expire soon, the nodes in the network will advise their neighbours to modify path and deliver data to another CH. Several nodes having the same number of neighbours play a vital role in ensuring that each node has a diverse number of neighbours. Each node delay time is estimated using where D t (i) is node i's delay time, E rem(i) is node i's residual energy, W t is primary CH's competition time, and R v is the random value. When nodes have same residual energy, random value can help to reduce communication conflicts. Algorithm for cluster setup phase (Algorithm 1):

Cluster Formation (CF).
e CF function chooses best non-CH candidate to join CH. After selecting principal CH, it will broadcast packet type 3 to node Si while waiting for the packet to arrive. If node Si receives a message from node Sj, add it to CH_list of potential nodes and alter it to a non-CH state. In this situation, it will calculate distance between non-CHs and CH to select best nodes as its CH. e best CH selection is determined by a small number of neighbour nodes, a short distance, and a high residual energy. Furthermore, more number of nodes in every cluster has an impact on the network's performance.

Transmission Phase.
is phase transfers data over the network between CHs and CMs. A CH rotation technique and stacking implementation make up the transmission phase. In the following part, we'll go over the CH rotation approach and the layered implementation design in more detail. After selecting the principal CH, the transmission procedure is used to create a TDMA schedule that transmits a broadcast schedule to all CMs. Data will be sent from the CMs to primary CH, which will aggregate it. Aggregation operations are forwarded to BS after that.

Analysing the Malicious Activities of the Network.
e network's nodes are responsible for not only delivering precise data, but also for updating new data in received data packet and sending it to next node. As a result, it is critical to confirm that data exchanged between nodes is secure. When a node is subjected to an aggressive or passive attack, it loses its trustworthiness. A carousal attack occurs when a rogue node causes a data packet to loop continuously, preventing it from reaching the BS or the destination node. e energy of the node depletes dramatically in a short amount of time as a result of this anomalous behaviour, resulting in a vampire attack. If BS does not receive requested data packet within specified time interval, BS requests that CH evaluate trust value of all its member nodes given by (2) T d (t) is computed direct trust between n1 and n2 nodes. P n1 (t) denotes packets that have been received. e total number of packets sent is P n2 (t). e surrounding nodes' estimated trust are represented in (3) Mathematical Problems in Engineering α, β values range from 0 to 1 such that α + β � 1. Trust threshold is 0.99-1. If estimated value falls below this threshold, sensor is classified as an MN, as it experiences higher packet loss during transmission. For effective transmission of data to send packet along chosen path, the trust criteria are taken into account. If BS identifies an MN based on the trust degree calculation after the verification procedure, it sends an alarm message to CH with ID and location of detected MN. CHs have now blacklisted and isolated rogue node, as well as broadcasting its ID to all of its members.

Authentication-Based Routing
Protocol. e sink is ready to receive a data packet from source/intermediate node k. It uses the selective authentication algorithm to determine whether or not to inspect the data packet. As a result, timer on the higher priority node is shorter. It would put up its own forwarder network and get ready to send data. e process is repeated by subsequent sensor nodes until data packet reaches sink.

Probabilistic Fuzzy Chain Set.
ree factors are chosen in the suggested work while considering the concept of vampire assault. ree trust factors, such as node's packet drop rate, percentage of battery drain, and the number of link requests initiated by the node, can be used to detect the vampire node. ese are described as follows.
(1) Packet Drop Rate (PDR): Malicious nodes lose a substantial percentage of packets, whereas trustworthy nodes forward all of the packets they receive. e ratio of number of packets dropped to total number of packets is used to calculate this value.
(2) Battery Discharge Rate (BDR): is node parameter is dependent on the network's operation. An active node appears to be in use and uses more energy. Nodes with a malignant or selfish aim, on the other hand, are observed to be more active than other nodes. (3) Number of link requests (NLR): e vampire node tries to connect to numerous nodes at the same time.
As a result, the node will generate an unusually large number of link requests. e ith node initial position is given by (5)   Mathematical Problems in Engineering e matrices' dimension is D. For example, the position of node 1 is given by x iD � x 1D � (x 11 , y 12 , z 13 , w 14 ), position of node 2 is given by Sx 2D � (x 21 , y 22 , z 23 , w 24 ), and so on. Fitness value is given by (6) f After determining the fitness of all nodes, the fitness with the lowest value, fbest, is regarded the best, and the fitness with the highest value, fworst, is measured worst. Masses of all surrounding nodes are determined in molecular dynamics. e atoms with the greatest mass have best fitness value, while atoms with smallest mass have poorest.
where M i (t) and M j (t) signify the mass of ith and jth atoms (node); fbest(t) indicates the least fitness value; and fworst(t) denotes the highest fitness value. PFCS is denoted as a collection of four tuples PFCS � (F H , F κ , F n , F d ) with various encouragement factors of vampire attack as FI � f1, fe, f3, . . ., fm. e following steps make it easier to mitigate a vampire attack with PFCS.

Mathematical Problems in Engineering 5
uncertainty factor' F n (χ g i,j ) and fuzzy associativity factor λF n(i) (f i ) as F n λ F n (J) (3) Finding of equivalent classes based on union and intersection operations utilising a probabilistic fuzzy theory technique, as shown in (2) and (3).
(4) Mean, higher, and lower degrees of approximations (π M , π G , π L ) evaluated for quantifying the degree of conformation that is estimated: Finally, the crisp factor of PFCS is calculated using (7) as the ratio of divergence between lower as well as greater approximations to increasing count of entities inspected in universe utilized for decision making.
(5) en evaluation of inclusion degree is enabled using  6 Mathematical Problems in Engineering output after applying the 27 rules and a high value is determined from the gathered fuzzy set. Probabilistic chain rule was used, which is the most often used due of its properties, to generate the eligibility index using Fuzzy Logic.
3.1.9. Defuzzification. Center of area (Z * ) method is utilized for defuzzification by (13) After calculating the eligibility index for all nodes, the threshold (TH) is derived using the equation (9).
For indiscriminate CH selection, every node in network creates a randomised number. If that value is less than computed TH, node will be assigned to CH role. CH's role is critical to the network's energy efficiency and rotated after each round to balance load among organized SN. Popt clusters are created in this manner. If some nodes remain after CF, they will join the cluster that is closest to them following CH recognition. Once all of the SN have been bound to clusters, the topology configuration process is complete. is method of weightage calculation enables for accurate values of criteria weights with the highest possible membership grade Input:n_round, deployment area, p, sing_p, initial_e, c_range, n_node, t_node Output:R_VALUE � [CHs,alive_n] (1) Initialization; (2) For r � 1 to n_round (3) For I � 1 to n (4) If(mod(r,3) � � 1) (5) If (S(i).energy ≤ 0) (6) Alive � alive-1; Mathematical Problems in Engineering for their membership function. Equation (10) is a linear programming model that enables the max-min weightage computation approach computationally.
e weightage set in question is made up of all strict and fuzzy sets with a separate high value intersected. e objective function is stated to be maximising in this scenario based on gain corresponding to result membership maximal grade. Furthermore, w i is the weight value of the crisp parameter linked with the jth condition.

Performance Analysis
e PFCS-ARP HC simulation configuration contains 100 mobile nodes that are randomly distributed within a terrain perimeter of 10001000 square metres. For validating the performance of PFCS-ARP HC, the pause period and simulation time were 20 seconds and 300 seconds, respectively.
e PFCS-ARP HC simulation environment uses 802.11 as MAC protocol with 2 Mbps channel capacity of and Constant Bit Rate data source. Source and destination pairs of 20 and 50 mobile nodes are also used in the simulation investigation. Table 1 shows comparative analysis for the proposed and existing techniques in minimizing the vampire attacks. Here the parameters compared are throughput, PDR, energy consumption, attack detection ratio, latency, and control overhead and compared with TAODV, LEACH, and WOA with the proposed PFCS-ARP_HC.    7 show comparative analysis of the proposed and existing techniques in vampire attack mitigation and improving the network performance. Comparative analysis has been carried out for throughput, PDR, energy consumption, attack detection ratio, latency, and control overhead. Here throughput attained by the proposed technique is 98% which enhanced when compared with TADOV, LEACH, and WOA. In terms of PDR, 89% has been obtained by the proposed technique; energy consumption by the proposed protocol is 75.5%, which is optimized when compared with existing protocol while data transmission. Since their energy consumption is high, the mitigation of vampire attacks by existing protocol is not efficient. Vampire attack detection ratio obtained by the proposed technique is 88% which is optimal than the existing techniques. Latency of the proposed technique in network is 46% and control overhead is 62% which is enhanced based on this comparative analysis.

Conclusion
is research offered a unique design strategy for minimising vampire attacks and improving network performance by utilising a security-based routing protocol and clustering-based data optimization. e data was optimised using a hybrid clustering approach, and the vampire attack was neutralised using a probabilistic fuzzy chain paired with an authentication-based routing protocol. e routing security of the network has been improved by authentication routing protocol. e performance can be improved by optimizing the data and network by hybrid-based sensor nodes clustering. Simulation for this technique has been carried out using NS2 and experimental results show that the performance of the proposed model in terms of throughput of 98%, packet delivery ratio of 89%, energy consumption of 67%, latency of 46% control overhead of 53%, and attack detection ratio of 87.9% [35].

Data Availability
e data used to support the findings of this study are available from the author upon request (kusumasyadav0@ gmail.com).

Conflicts of Interest
e authors declare that they have no conflicts of interest. Mathematical Problems in Engineering 9