Dwarf Mongoose Optimization-Based Secure Clustering with Routing Technique in Internet of Drones

: Over the last few years, unmanned aerial vehicles (UAV), also called drones, have attracted considerable interest in the academic ﬁeld and exploration in the research ﬁeld of wireless sensor networks (WSN). Furthermore, the application of drones aided operations related to the agriculture industry, smart Internet of things (IoT), and military support. Now, the usage of drone-based IoT, also called Internet of drones (IoD), and their techniques and design challenges are being investigated by researchers globally. Clustering and routing aid to maximize the throughput, reducing routing, and overhead, and making the network more scalable. Since the cluster network used in a UAV adopts an open transmission method, it exposes a large surface to adversaries that pose considerable network security problems to drone technology. This study develops a new dwarf mongoose optimization-based secure clustering with a multi-hop routing scheme (DMOSC-MHRS) in the IoD environment. The goal of the DMOSC-MHRS technique involves the selection of cluster heads (CH) and optimal routes to a destination. In the presented DMOSC-MHRS technique, a new DMOSC technique is utilized to choose CHs and create clusters. A ﬁtness function involving trust as a major factor is included to accomplish security. Besides, the DMOSC-MHRS technique designs a wild horse optimization-based multi-hop routing (WHOMHR) scheme for the optimal route selection process. To demonstrate the enhanced performance of the DMOSC-MHRS model, a comprehensive experimental assessment is made. An extensive comparison study demonstrates the better performance of the DMOSC-MHRS model over other approaches.


Introduction
With the growth of modern wireless communication technology, the Internet of things (IoT) is becoming a broadly utilized technology in the domain of several intellectual applications and services [1]. Eventually, the rise in interconnectivity between numerous things or objects has produced massive data. But, such IoT applications were not as much

Existing Works on Cluster-Based Routing in IoD
In [11], the operation of drones in ad hoc mode and their cooperation with vehicles in vehicular ad hoc networks (VANET) were learned to aid the detection and routing procedure of malicious vehicles. A routing protocol termed vehicular routing unit (VRU) can be devised that involves two different ways they are routing packets of data between UAVs using a protocol termed VRU_u and supplying packets of data between vehicles by using drones utilizing a protocol called VRU_vu. In [12], a novel systematic structure can be presented for solving the issue of multi-drone collaborative task allotment. It can be developed as a combinatorial optimizing issue and resolved by the enhanced clustering technique. The main goal was to enable multi-drone for completing tasks has less energy utilization. Since the drone count increased, it appears that flight safety problems such as collisions between the drones, and an enhanced multi-UAV collision-resistant approach related to the enhanced artificial potential domain were devised. Namdev et al. [13] modeled a whale optimization algorithm-related optimized link state routing (WOA-OLSR) over a flying ad hoc network (FANET) for providing optimum routing for secure and energy-efficient FANET. The efficacy of OLSR can be improvised with the help of WOA and assessed performance displays superior efficacy of WOA-OLSR. In [14], the optimal CH selection depends upon blockchain (BC) transactions, residual energy (RE), mobility, online duration, connectivity, and reputation by utilizing improved artificial bee colony optimization (IABC). The presented IABC uses two distinct search equations for onlooker bee and employee bee for enhancing exploitation abilities and convergence rate. In addition, a lightweight BC consensus technique, the AI-proof of witness consensus algorithm (AI-PoWCA) can be projected that employs the optimal CH for mining. In [15], drones that can act as mobile sinks were taken into account and prevailing work on wireless sensor network (WSN)-UAV atmosphere authentication was protracted. A secure authentication structure utilizing an elliptic-curve crypto-system was provided. The projected structure can be assessed to assure it is resilient to renowned potential assaults relevant to password guessing, data confidentiality, key impersonation, and mutual authentication.
In [16], a secure and reliable routing protocol (SecRIP) for the FANET can be devised for reliable and efficient data communication. This script operates toward the improvement of the quality of experience (QoE) metrics and quality of service (QoS). The script operates on two methods: the dragonfly technique and the chaotic algae technique; such methods serve the functionalities of cluster management, selection, and data communication in intercluster. Khan et al. [17] project a new routing approach as the extension of AntHocNet because of mobile features; it is called a flying nature-inspired method. Moreover, a case study was performed for improving the signal power with the help of modeled learning technique named decision tree (DT).

Paper Contributions
This study develops a new dwarf mongoose optimization-based secure clustering with a multi-hop routing scheme (DMOSC-MHRS) in the IoD environment. The goal of the DMOSC-MHRS technique involves the selection of CHs and optimal routes to a destination. In the presented DMOSC-MHRS technique, a new DMOSC technique is utilized to choose CHs and create clusters. A fitness function (FF) involving trust as a major factor is included to accomplish security. Besides, the DMOSC-MHRS technique designs a wild horse optimization-based multi-hop routing (WHOMHR) scheme for the optimal route selection process. To demonstrate the enhanced performance of the DMOSC-MHRS model, a comprehensive experimental assessment is made.

Paper Organization
The organization of the paper is given as follows. Section 2 introduces the proposed DMOSC-MHRS model and the experimental analysis of the DMOSC-MHRS model is provided in Section 3. Lastly, Section 4 concludes the study with major findings and possible future enhancements.

The Proposed Secure Clustering with Routing Protocol
In this study, a new DMOSC-MHRS technique has been developed for secure clusterbased communication in the IoD environment. The DMOSC-MHRS technique proficiently chooses CHs and optimal routes to a destination. The major intention of the proposed model is to accomplish security, energy efficiency, and improved lifetime. Figure 1 showcases the overall procedure of the DMOSC-MHRS algorithm.

Overview of DMO Algorithm
The DMO technique inspires the performance of dwarf mongooses when determining their food [18]. Generally, the DMO starts with setting the primary value to a group of solutions utilizing the subsequent equation: whereas refers to the arbitrary number created in zero and one. and implies the restrictions of the searching area. The swarm of DMO has three groups such as babysitters, alpha group, and scouts. All the groups have their individual performance for capturing the food, and the particulars of these groups are provided as: The fitness of all the solutions is calculated if the population was established. Equation (2) computes the possible value for all the fitness populations, and alpha female ( ) is selective and dependent on this probability.

=
( 2) relates to the number of mongooses from the alpha group. The number of babysitters was represented by bs. Peep is the vocalization of the dominant female which keeps the family on track.
All the mongooses sleep from the primary sleeping mound that is fixed ∅. The DMO utilized for generating a candidate food place.
The sleeping mound was offered in Equation (4) then all the repetitions, whereas ℎ signifies the uniformly distributed arbitrary value in −1 and 1.
Equation (5) comprises the average value of sleeping mounds.

= (5)
If the babysitting alters condition is met, this technology advances to the scouting phase, in which the next sleeping mound or food source is assumed.

Scout Group
As mongooses are identified to not go back to past sleep mounds, the scout's appearance is for the next sleeping mounds, making sure to search. In this method, scout and

Overview of DMO Algorithm
The DMO technique inspires the performance of dwarf mongooses when determining their food [18]. Generally, the DMO starts with setting the primary value to a group of solutions utilizing the subsequent equation: whereas rand refers to the arbitrary number created in zero and one. u j and l j implies the restrictions of the searching area. The swarm of DMO has three groups such as babysitters, alpha group, and scouts. All the groups have their individual performance for capturing the food, and the particulars of these groups are provided as:

Alpha Group
The fitness of all the solutions is calculated if the population was established. Equation (2) computes the possible value for all the fitness populations, and alpha female (α) is selective and dependent on this probability.
n relates to the number of mongooses from the alpha group. The number of babysitters was represented by bs. Peep is the vocalization of the dominant female which keeps the family on track.
All the mongooses sleep from the primary sleeping mound that is fixed ∅. The DMO utilized for generating a candidate food place.
The sleeping mound was offered in Equation (4) then all the repetitions, whereas phi signifies the uniformly distributed arbitrary value in −1 and 1.
Equation (5) comprises the average value of sleeping mounds.
If the babysitting alters condition is met, this technology advances to the scouting phase, in which the next sleeping mound or food source is assumed. As mongooses are identified to not go back to past sleep mounds, the scout's appearance is for the next sleeping mounds, making sure to search. In this method, scout and forage were carried out concurrently. This drive was exhibited then a successful/unsuccessful search for a novel sleeping mound. Specifically, the migration of mongooses is contingent on their entire efficiency. The scout mongoose is defined in Equation (12).
In which rand demonstrates a random number from the range of zero and one, Max iter ) whereas the parameter which regulates the mongoose group, the collective-volitive movement was reduced linearly as the iterations developed.
in which the mongoose's movement to the novel sleeping mound was defined by this vector.

Babysitters Group
The babysitters were commonly inferior group members which continue with the young and cycle on a regular basis allowing the alpha female (mother) for leading the rest of the group on daily forage expeditions. The number of babysitters was proportional to the size of the population; it can be a stimulus for the technique by decreasing the entire population size dependent upon the percentage set. The scout and food source data earlier indicated by the family members replaces them by resetting the use of the babysitter interchange parameter.

Design of DMOSC Technique
In the presented DMOSC-MHRS technique, a new DMOSC technique is utilized to choose CHs and create clusters. The FF in this MOTAHO is used for choosing the topical CH derived. Now, the FF is expressed by the four dissimilar parameters namely number of hops, trust, distance, and residual energy (RE) [19].
Trust: In CH selection, trust is regarded as a key parameter in the FF to improve security. The mutual trust made in a specific period is used for accomplishing the transmission.
Direct trust (DT) is predicated on the approximate period of communication among i th node and d th destination n. DT is measured as the gap on the list of actual and the projected period of i th node for authenticating the public key expressed by the d th destination. Hence, DT including i th node and d th the destination is given by, In Equation (7), τ appx defines the approximate period and τ est determines the estimated period to authenticate the public keys. In other words, τ appx and τ est are the expected period for sending and receiving the public keys via the destination and also the node. ω implies the opinion parameter of this node.
The node with the opinion parameter is plotted based on DT. However, the node without a witness parameter is authenticated by the indirect trust (IDT) as follows that is given by, In Equation (8), r denotes the total neighbors of node i. Recent trust (RT) is measured by the DT and IDT along with the crucial validity and admits the destination or sink that is given in part of the moment.
Distance: It determines the distance (g 2 ) amongst the CH to the BS and the next-hop node. Since the energy usage of nodes is proportionate to the distance of the communication path. Consequently, it is essential to determine the communication path with a lesser distance for diminishing energy utilization.
Residual energy: The candidate CH with higher RE (g 3 ) formulated in Equation (11) is greatly desirable at the time of CH selection. Since the CH has to perform different operations namely data transmission, collection, and aggregation.
In the above equation, E CH i illustrates the RE of CH. Some hops: Some standard nodes belonging to the specific CH are described by some hops. The energy utilization of CH is lesser when it has a lesser number of hops. Therefore, the CH with lesser hops is regarded in the number of hops (g 4 ) and CH selection is formulated as follows.
In Equation (12), the number of standard nodes for the specific CH can be represented as I i . The mentioned objective values were converted into one objective related to the weighted sum technique as presented below in Equation (13).
where the δ 1 , δ 2 , δ 3 , and δ 4 represents the weights allotted to every FF value. The devised FF was utilized from the DMOSC approach to choose an optimum CH.

Process Involved in WHOMHR Technique
In this study, the DMOSC-MHRS technique designs a WHOMHR scheme for optimal route selection process. The WHO algorithm is a metaheuristic algorithm dependent upon the social life of wild horses [20]. During this technique, distinct performances are demonstrated by wild horses namely leading, chasing, grazing, hunting, and mating. The horses were categorized into two social groups such as territorial and non-territorial. But, the WHO technique concentrates on the non-territorial group that comprises group leaders named stallions, several mares, and their offspring. The part of stallion is to lead the group and connect with mares, as the foals start their lives with grazing performance. In addition, if the foals exceed the age of puberty, they can leave their group and integrate into other groups. The process of the WHO technique was outlined in the subsequent steps:

Population Initialization
During this stage, the parameter needed for the WHO technique was established for evaluating the primary solution, afterward upgraded based on the technique process. The horses were separated into many groups and all the groups have one stallion. This division was estimated employing in Equation (14) as follows: where H denotes the entire amount of groups, Q refers to the population size, and SR signifies the number of stallions from the population.

Grazing Behavior
This step presents the grazing performance of foals before they can obtain puberty. The stallion was considered at the center of the grazing region, whereas the residual group members were adjacent to the center of the region. This performance is demonstrated employing in Equation (15): In which i signifies the number of group members, j denotes the number of stallions, X j ι+1,H , X j i,H stands for the place of group members from the next and present iteration correspondingly, A has an arbitrarily selective adaptive process, R represents the arbitrary number in −2 and 2, and S j denotes the stallion place.

Horse Mating Behavior
This phase offers the performance of foals afterward obtaining puberty age. As previously noted, foals leave their groups and combine with another group for mating and for preventing fathers from marrying their daughters and sisters. Besides, this performance was demonstrated in employing Equation (16): where X t H,l signifies the place of horse t of group l, X u H,i refers to the place of foals u of group i, and X w H,j signifies the place of foal w of group j, in which the foal u mate with foal w from the group l. Therefore, an essential state to mate was obtained. Algorithm 1 demonstrates the working of the WHO algorithm.

Group Leadership
During this stage, the group stallion leads the members of the group to the waterhole for food. Likewise, the stallion fights with another stallion for dominating the waterhole. This performance is defined utilizing Equation (17): In which S ι+1,G , S ι,G demonstrates the next and present place of leaders correspondingly, WP implies the place of waterholes and r 1 is a random vector among zero and one.

Leaders' Exchange and Selection
At last, the group leader was chosen for obtaining an optimum fitness value. During all the iterations, the group leader was selected, whereas an optimum leader is obtained amongst the entire leaders from the iterations. This step is demonstrated by Equation (18): During this work, the ending condition is for performing the optimized procedure up to the maximal count of iterations (Max. It). The optimized approach was computed employing 100 iterations, with a population size of 30. An important objective of the WHOMHR approach is maximizing network lifespan and minimizing energy consumption of all drones [21]. Assuming that h1 is a most main function such that CHs select next-hop CHs with superior RE to route the data such that for maximizing the network lifespan viz., h1 is maximized. Consider h2 to be another objective function that is minimal distance amongst CHs to next-hop CHs and next-hop CHs to BS. Decreasing the energy consumption of networks requires minimizing the h2. Assume that h3 is the 3rd objective function such that CHs are select as the next-hop CHs with lesser node degree. To enhance the network, lifespan requires minimizing h 3 . Let b ij be a Boolean variable determined as: subject to, Drones 2022, 6, 247 9 of 16

Results and Discussion
In this section, the secure communication performance of the DMOSC-MHRS model is investigated in detail. The proposed model is simulated using MATLAB under three different scenarios based on grid size.       Table 2 presents an overall energy consumption (ECM) inspection of the DMOSC-MHRS algorithm with recent models on different scenarios. Figure 3 reports a comparative ECM assessment of the DMOSC-MHRS technique with existing models on Scenario-1.  Table 2 presents an overall energy consumption (ECM) inspection of the DMOSC-MHRS algorithm with recent models on different scenarios. Figure 3 reports a comparative ECM assessment of the DMOSC-MHRS technique with existing models on Scenario-1.         Table 3. Figure 6  A comprehensive cluster lifetime (CLT) inspection of the DMOSC-MHRS model with other models is performed on different scenarios in Table 3. Figure 6

Conclusions
In this study, a new DMOSC-MHRS algorithm was devised for secure cluster-based communication in the IoD environment. The DMOSC-MHRS technique proficiently chooses CHs and optimal routes to a destination. In the presented DMOSC-MHRS approach, a new DMOSC technique is utilized to choose CHs and create clusters. A FF involving trust as a major factor is included to accomplish security. Besides, the DMOSC-MHRS technique designs a WHOMHR scheme for the optimal route selection process. To demonstrate the enhanced performance of the DMOSC-MHRS model, a comprehensive experimental assessment is made. Extensive comparison studies illustrate the better performance of the DMOSC-MHRS model over other approaches, with maximum reliability of 96.99%. Therefore, the proposed model can be employed in future real-time applications such as environmental monitoring, forest fire detection, disaster management, search and rescue, and smart cities. In the future, the performance of the DMOSC-MHRS algorithm can be enhanced by the utilization of data aggregation approaches, thereby enhancing overall network efficiency. In addition, the overall network performance can be improved by the use of unequal clustering techniques to mitigate hot spot problems.

Conclusions
In this study, a new DMOSC-MHRS algorithm was devised for secure cluster-based communication in the IoD environment. The DMOSC-MHRS technique proficiently chooses CHs and optimal routes to a destination. In the presented DMOSC-MHRS approach, a new DMOSC technique is utilized to choose CHs and create clusters. A FF involving trust as a major factor is included to accomplish security. Besides, the DMOSC-MHRS technique designs a WHOMHR scheme for the optimal route selection process. To demonstrate the enhanced performance of the DMOSC-MHRS model, a comprehensive experimental assessment is made. Extensive comparison studies illustrate the better performance of the DMOSC-MHRS model over other approaches, with maximum reliability of 96.99%. Therefore, the proposed model can be employed in future real-time applications such as environmental monitoring, forest fire detection, disaster management, search and rescue, and smart cities. In the future, the performance of the DMOSC-MHRS algorithm can be enhanced by the utilization of data aggregation approaches, thereby enhancing overall network efficiency. In addition, the overall network performance can be improved by the use of unequal clustering techniques to mitigate hot spot problems.  Data Availability Statement: Data sharing is not applicable to this article as no datasets were generated during the current study.