Metaheuristic Quantum Glowworm Swarm Optimization based Clustering with Secure Routing Protocol for Mobile Adhoc Networks

Mobile adhoc network (MANETs) comprises a collection of independent, compact sized, and inexpensive sensor nodes, which are commonly used to sense the physical parameters in the geographical location and transmit it to the base station (BS). Since clustering and routing are considered as the commonly used energy efficient techniques, several metaheuristic algorithms have been employed to determine optimal cluster heads (CHs) and routes to destination. But most of the metaheuristic techniques have failed to achieve effective clustering and routing solutions in large search space and the chance of generating optimal solutions is also considerably reduced. To resolve these issues, this paper presents a new Metaheuristic Quantum Glowworm Swarm Optimization based Clustering with Secure Routing Protocol for MANET, named QGSOC-SRP. The presented QGSOC-SRP technique follows two stage processes, namely optimal CH selection and route selection. Firstly, the QGSO algorithm derives a fitness function using four variables such as energy, distance, node degree, and trust factor for optimal election of secure CHs. Secondly, the SRP using oppositional gravitational search algorithm (OGSA) is applied for the optimal selection of routes to BS. The traditional GSA is inspired by the law of gravity and interaction among masses. To improve the effectiveness of the GSA, OGSA is derived based on the oppositional based learning concept for population initialization and generation jumping. For validating the effective results of the presented OGSOC-SRP technique, a set of experiments were performed and the results are determined interms of distinct measures.


Introduction
In general, Mobile Ad Hoc Network (MANET) is a multi-hop transitory as well as selfsufficient system configured by mobile nodes and wireless connections [1]. In MANET, a node is capable to join and relieve from a system dynamically [2]. Due to the scalable and infrastructure less behavior of MANET, it is employed in military services and alternate civilian applications. The key objective of MANET is to exchange information among rescuers, searching teams, and clinical professionals, especially in emergency sites in case of complete network failure [3]. Also, various other habitats in routine life such as weather forecasting, ocean tracking, traveler location, destination monitoring, and social network [4].
Using the closer range space and rapid deployment of airliner like drones, MANETs are composed of high diversity. But, MANETs are limited with insufficient speed limit of battery, and energy which is still a complicated problem for various applications [5]. For inadequate power of MANET, developing a routing protocol is to ensure the extensive convergence, and improve the efficiency of novel routes. Additionally, it is significant to enhance the power utilization and make sure an effective data transmission, and which modifies the aggressive changes processed by network topology.
Numerous traditional models have been employed on energy equalizing protocols. Usage of effective energy mechanism intends to mitigate the power application and improvise the system duration. Typically, the protocols are divided into 2 categories: low energy and maximum network lifetime routing protocols. First, routing protocols with low power identifies a power effective routing path from source to target [6] while improvised network lifetime manages to balance the node power and explore energy-saving router [7].
Additionally, ant colony mechanism performs well by means of power management and is extensively used in multiple domains. Next, a bionic concept mimics the character of living organisms on Earth. The major objective of this method is to achieve optimal results with minimum power resources. Mostly, it is employed to enclose simple heuristic approaches within dispersed tasks [8]. The Ant Colony Optimization (ACO) method contains massive benefits and features. Actually, ACO is a new heuristic approach used for resolving integrated optimizing problems. Few characteristics of ACO are positive feedback, distributed processing, and dynamic changes [9]. Baras and Mehta [10] deployed Probabilistic Emergent Routing Algorithm (PERA). The extended version of ACO for MANET is referred as PACONET as defined in [11] applies forward and backward ants to increase the pheromone level that is different from AntNet concept. Initially, forward ant enhances pheromone to maximize the convergence of a system and results in trapping within local optimum issues. It is composed of maximum broadcast as each ant needs time for discovering path and ID of previous nodes. In order to overcome these problems, a basic ACO gets easily trapped within the local optimum, and Bullnheimer et al. [12] presented a Rank-Based Version of Ant System (ASrank). It is projected by developing a mechanism of sorting in Genetic Algorithm (GA).
Woungang et al. [13] established ant-swarm based energy-effective ad hoc on-demand routing protocol (ACO-EEAODR) by means of RE and router length of a node. Here, weighting factors of 2 characteristics in a protocol gain various metrics according to different cases, whereas pheromones of each node are expanded over residual energy (RE). In de Figueiredo Marques et al. [14], ACO is utilized in mesh routing system to identify best routes while managing shared coordination among nodes with Low-Power and Lossy network (LNNs) with minimum energy application. Now, ACO is named LNNACO which enhances the transmission rate by considering the power utilization in Max-Min-Path(MMP) ACO method. Since the LNNACO method is unable to identify the effect of maximum mobility and it is unfit for high speed networks.
In Zhou et al. [15], AC based Energy Control Routing Protocol (ACECR) has been presented.
It considers maximum and minimum RE for each path. For route discovery, backward ant upgrading is enabled and pheromone table is based on minimum power and hop count.
However, ACECR is unable to make sure the complete transmission energy can be limited with defined router. Mohsen [16] utilized Simulated Annealing (SA) concept for identifying global optimal path, by the combination of SA and ACO. Some of the existing models are time consuming by means of computing the routing path and distance of a router is maximum when compared with ACO approach.

This paper introduces a novel Metaheuristic Quantum Glowworm Swarm Optimization based
Clustering with Secure Routing Protocol for MANET, named QGSOC-SRP. The presented QGSOC-SRP technique involves OGSO based optimal CH selection and SRP based route selection. The SRP using oppositional gravitational search algorithm (OGSA) is applied for the optimal selection of routes to BS. To improve the effectiveness of the GSA, OGSA is derived based on the oppositional based learning concept for population initialization and generation jumping. The experimental validation of the presented model takes place in different aspects.

The Proposed QGSOC-SRP technique
The workflow involved in the presented QGSOC-SRP method is depicted in Fig. 1. Consider a MANET with 'n' mobile sensor placed randomly wherever required. A sensor node is deployed and begins to collect information reading the surrounding environment. Then, the QGSOC algorithm gets executed by the Base station (BS) to elect the cluster heads (CHs) by the use of a fitness function (FF) involving four variables such as energy, distance, node degree (ND), and trust factor for optimal election of secure CHs. Followed by, OGSA based routing technique is applied to pick up an optimal path to BS. Finally, the data will be transmitted from cluster members (CMs) to CHs and thereby to BS.

QGSOC Technique
Here, the detailed performance of QGSO algorithm based clustering process and the optimal election of CHs are provided clearly.

Glowworm swarm optimization (GSO)
Usually, GSO is considered as modern swarm optimization method evolved from the luminescent feature of fireflies. Here, the glowworm swarms are distributed in a solution space as well as fluorescence intensities are relevant to the FF of glowworm's location. The position of a glowworm depends upon the intensity of brightness which means the FF values are maximum. Also, glowworms have a dynamic line of vision, which is referred as decision domain, where the range is relevant to density of adjacent nodes. Once the maximum number of iterations is reached, the glowworms are placed in best locations [17]. Typically, this method is composed of 5 phases namely, upgrading the fluorescein concentration, increment in neighbor set, enhance the decision domain radius, update moving probability, and upgrade glowworm position. Hence, fluorescein concentration increment mechanism is classified by where (t) implies fluorescein concentration of ℎ glowworm at time f, means the fluorescein volatilization coefficient, denotes the fluorescein enhancing factor, ( ) refers the FF value and (t) defines the location of glowworm at time. Therefore, increment in neighbor set is defined by where (t) refers the neighbor set of ℎ glowworm at time t as well as d ( ) mimics the radius of decision domain of ℎ glowworm at moment . Thus, enhancing decision domain radius mechanism is illustrated in Eq. (3).
Where r denotes the perceived radius of glowworm, implies the rate of change of decision domain, and signifies the adjacent threshold. As a result, update moving probability of a framework is demonstrated in Eq. (4).
Where (t) implies the probability with glowworm moves to glowworm at time t. The upgrade glowworm position mechanism is depicted in Eq. (5).
For enhancing the working function of GSO method, quantum computing is incorporated. Fig. 2 illustrates the flowchart of GSO model.

Quantum GSO Algorithm
Quantum computing is considered as a computation model which employs the relevant mechanism to quantum theory like quantum estimation, state superposition, as well as quantum entanglement. The basic unit of quantum processing is qubit. The basic states |0 > and |1 >develops a qubit which is illustrated as linear combination of 2 fundamental states as shown in the following, | | 2 defines the possibility of observing state |0 >, | | 2 implies the possibility of observing state |1 >, in which | | 2 + | | 2 = 1. A quantum is deployed with the help of qubits. Due to the quantum superposition, a quantum is comprised of 2 possible metrics. Finally, n-qubits quantum is demonstrated as provided in the following.
The condition of qubits like rotation gate, NOT gate, Hadamard gate,are modified using quantum gtaes. Initially, rotation gate [18] is defined as a mutation operator used for developing quanta mechanism and best solutions to identify an global optimal solution.
The rotation gate is illustrated as follows: refers the size and dimension of rotation respectively.

QGSO based CH Selection Process
The FF of QGSO is applicable in resolving the optimal clustering issues. Due to the complicated nature, maximum control data has to be replaced between the nodes in CH election, which results in system overhead. In order to overcome this issue, QGSO FF has been applied. Also, the local density of CH and average distance inside a cluster, power utilization of nodes, and dispensability of CH are some of the parameters used in controlling the generation of unequal network clustering.
In CH election phase, distribute CH, eliminate missing data, develop adjacent node to join CH and power utilization is higher than CM nodes. With no balancing metrics, the premature death of CH occurs and power drain rapidly. The major node in a system is assumed to be CH. Thus, the position of CH has to be determined and to reduce the size of CH nearby sink or BS nodes, where multiple CH accounts for data forwarding process and increase the realtime process and power utilization of CH.
Consider N nodes for developing K clusters with M(K << ) candidate CH. Followed by, a network is composed of C clustering models and select best clustering mechanism are referred as optimization issues. Under the application of GSO FF, the optimal clustering issues can be resolved and FF assumes local density of CH, average distance inside a cluster, and energy dispersion of nodes.
Assume a WSN with N sensor nodes and develop K clusters [19]. The CH election is performed by using QGSO FF to resolve the clustering issues. Hence, FF is obtained using 5 input attributes like RE, communication expense, trust degree, ND as well as node marginality.

Residual energy
It is a significant parameter applied in WSN. Here, CHs need high power than CMS because of processing aggregation, computation, and data routing. Therefore, RE is evaluated by using applied expression: where 0 and denotes the basic energy and power used by nodes, respectively; and indicates the RE of regular node.

Communication cost
The forwarding data has observed energy which is directly proportional to square of distance among 2 nodes. The count of processing cost is demonstrated as given below: Where signifies the high distance between a node and neighbors; 0 refers the broadcasting radius of a node.

Trust degree
Basically, a node is assigned to triangle function (TF) value 1. Next, value of TF is restricted by independent prediction concept whereas a node process anomalous operation and node is decided as anxious node and TF nodes are demonstrated in Eq. (12): Here, nodes with ΤF modifies the opportunity of becoming CHs. When the node falls under ΤF of malicious node, it is removed from cluster and suspicious nodes are considered as CM and not as CH.

Node degree
The rule applied in ND is that nearby neighbors model has optimal measure; a node with maximum probability is declared as CH. Hence, ND is computed as given below: Where, defines the measure of neighboring nodes and 0 implies the optimal number of neighbors.

OGSA based SRP
Once the CHs are chosen, the subsequent process is to elect an optimal set of CHs. It is performed by the use of OGSA. The detailed working of the OGSA based route selection process is defined here.

Overview of GSA
Generally, GSA [20] is composed of agents which are assumed as objects and performances In GSA is mass or agent is composed of 4 specifications namely, position, inertial mass, active and passive gravitational mass. Initially, position of a mass refers a solution to the problem, and gravitational as well as inertial masses can be estimated under the application of FF value. Besides, a mass is considered to be a solution. Also, it is directed by an appropriate modification of gravitational, and inertial masses. As a result, optimal solution might be gained in a search space. Mostly, the given rules are followed by the masses.
i. Law of gravity: A particle attracts the adjacent particles by the existence of gravitational force among 2 particles are directly proportional to product of masses and inversely proportional to square of distance (R). R is employed as an alternative to 2 since R has provided considerable outcome when compared with 2 with standard test functions.
ii. Law of motion: The recent velocity of a mass is similar to summation of fractions from existing velocities and differences in the velocity.
Then, assume a model with agents. Also, location of ith agent is depicted by: Where denotes the location of th agent in th dimension. At time , force employed on th mass from th mass is described as shown the following expression, where ( ) and ( ) depicts the active as well as passive gravitational mass relevant to th and th agent at time , ( ) shows a gravitational constant at time , refers a minimum constant, and ( ) signifies the Euclidian distance among 2 agents and j as expressed in the following, In order to provide a stochastic feature, it is desired that the overall force applied on th agent in th dimension is randomly weighted sum of th elements of forces released from alternate agents represented by given function: where denotes a random value from [0,1]. Therefore, using law of motion, the simulation of th agent at time in th dimension is illustrated by the given expression: where int ( ) depicts the inertial mass of th agent. The position and velocity of an agent could be measured by using given expression.
In (19), implies an equal random variable from where 0 is allocated to 100, is declared to 20, and max demonstrates the recent and overall count of iterations correspondingly. Gravitational and inertia masses are determined using the fitness estimation. By considering the uniformity of gravitational mass and inertia mass, the scores of masses are determined with the help of fitness mapping and it is upgraded by given expressions: In (28), refers the group of agents with optimal fitness measures and heavier masses.

Opposition-based learning
Evolutionary optimization models initiates few solutions and attempts to maximize the supremacy of optimal solution(s). The process of exploration is terminated if predefined condition is fulfilled. Under the inexistence of apriori data, it is often initialized with random guesses. Hence, processing time is related to distance of initial guesses from optimal solution.
The chance of simulating fitter solution is improved by validating the opposite solution [21].

Opposition-based gravitational search algorithm
Alike the population-based optimization approaches, 2 major steps are differentiable for GSA, like population initialization and make new generations using GSA. Recently, the principle of OBL is embedded in 2 phases [22]. The actual GSA is selected as parent model and opposition-based procedures are incorporated with the aid of implementing simulated convergence profile.

QGSA based Route Selection Process
Here, the newly developed OGSA based routing method has been defined for computing efficient routing process applies a heuristic mechanism. When compared with other heuristic models, this approach provides optimal solution for identifying optimal route for reducing power consumption.
In this framework, system with 'N' count of CHs are assumed and ith CH place is depicted by Eq. (31). At time t, force applied on ith CH from jth CH along with static mass is demonstrated in Eq. (32).
where µ denotes the simple constant, g(t) indicates the gravitational force at time 't', and ( ) refers Euclidean distance from node 'i' and node 'j'.

( ) = ‖ ( ), ( )‖ 2 (33)
The overall force employed on ith CH exerted from alternate CH with d dimension is assumed to be weighted sum of force acting from adjacent CH is depicted in Eq. (34).
Based on the law of motion, acceleration is triggered by force. If mass is constant, then the acceleration is presented for overall force applied on ith CH to 'jth' CH. In this model, acceleration of CH 'i' at 't' time is presented in Eq. (35).
The CH election is carried out randomly. Hence, velocity and position of CH should be determined [23]. Also, velocity of CH at time t + 1 is processed by the recent velocity along with the acceleration. Therefore, upcoming position of 'ith' CH and corresponding velocity is evaluated by Eqs. (36) and (37).
It is used in finding the optimal route between alternate routes. When best route is not found then follow the same process until reaching the termination criteria. Table 1

Conclusion
This paper has developed an effective QGSOC-SRP technique to achieve energy efficiency and prolong network lifetime in MANET. The presented QGSOC-SRP technique follows two-stage processes, namely optimal CH selection and route selection. The sensor nodes are initialized primarily and collect information about the surrounding environment. Then, the QGSOC algorithm gets executed by the BS to elect the CHs by the use of a FF involving four variables such as energy, distance, ND, and trust factor for the optimal election of secure CHs. Followed by, OGSA based routing technique is applied to pick up an optimal path to BS. Finally, the data will be transmitted from CMs to CHs and thereby to BS. To improve the effectiveness of the GSA, OGSA is derived based on the oppositional based learning concept for population initialization and generation jumping. The experimental validation of the presented model takes place in different aspects. The experimental outcome justified the superior performance of the presented model over the compared methods.
Funding Statement: The authors received no specific funding for this study.

Conflicts of Interest:
The authors declare that they have no conflicts of interest to report regarding the present study.