Intelligent energy-aware multiple quality of service restraints based secured optimal routing protocol with dynamic mobility estimation for wireless sensor networks

Abstract

This research work confronts a sender-based responsive and novel protocol named “Intelligent Energy-Aware Multiple restraints Secured Optimal Routing (IEAMSOR)” protocol for WSNs. In order to deal with the various emerges like packet routing, node mobility, and energy optimization as well as energy balancing in WSNs. The proposed protocol accounts for the basic QoS restraints such as Delay, HopCount and Energy Level for each link of ‘n’ number of routes and predicts the best optimal path among these in-between sender and receiver nodes throughout the route discovery process. It also assures the energy level of each node existing on the route during the route reply process. It incorporates the modified mobility prediction approach in order to estimate the stableness of link failure time for every link of each path during the route reply process. The main objective of this work to achieve the energy balancing among the nodes is achieved through fuzzy rules based node’s trust classification is introduced and based on this energy weight of each node is adjusted according to their trustworthiness. It accomplishes the path sustainment process when the link among the two nodes goes down. Moreover, the proposed model has been given careful attention for selecting additional substitute routes throughout link failure. The experimental results have seemed that the IEAMSOR protocol performs better than the existing traditional protocols.

Keywords

QoS energy level fuzzy rules mobility prediction optimal path trustworthiness

1 Introduction

Sensor networks are often changing good environments that may monitor close conditions like temperature, movement, sound, light, location and etc. Wireless Sensor Networks (WSNs) technology presents its distinctive style challenges. One vital feature that distinguishes sensor networks from ancient distributed systems is would like for energy potency [1, 2]. Several nodes in rising systems would like to save energy from an electric battery. The size of a sensing element network readying can build recharging these energy reserves not possible. Though energy potency is often improved at varied layers of the communication protocol stack, most revealed analysis has targeted on hardware-related energy potency aspects of wireless communications. Low-power electronic components, power-down modes and energy economical modulation area unit samples of add this class. However, as a result of elementary physical limitations, progress towards additional energy potency is anticipated to be achieved in alternative layers.

A wireless sensing element network comprises of an oversized range of unattended, typically self-organized small sensors, of the size of the order of a cubic centimeter, scattered in a section for a selected application. Every small sensing element is capable of sensing information from the setting, activity easy computations and transmission of this information over wireless medium either on to Base center or through some cluster head, usually referred to as gateway [3]. WSNs though have some similarities with ad-hoc networks however they take issue from ad-hoc networks principally because of their additional serious energy constraints, a lot of larger density of sensing element nodes, lower price and typically static nature of nodes [4]. Moreover, the WSNs area unit designed for the military operation, instead of distributed computing.

Sensors nodes are battery operated and once deployed are unattended and expected to control for an extended amount of lifetime, typically from some months to years. Thus, energy may barely be a resource in a very wireless device network and thus its economical usage is crucial for extending the lifetime of the total sensor device network. A sensor’s energy is especially consumed by the following major actions such as detection, calculating and communication.

Energy consumption in sensing element networks maybe a hanging issue, it’s owing to that the batteries carried by every mobile node have restricted power offer, process power is restricted, and that successively limits the standard and amount of services and applications that may be supported by every node. In sensing element networks, the nodes play parallel job responsibilities like information providing, processing, and information routing. The energy needed for information sensing varies to the type of application.

Chung et al. [5] suggested a packet route excerption technique to enhance the sensing power based on the statistical filtering approach. Every node measures the sensing power of each incoming route from the base station (BS) and selects the foremost trusted route for information packet delivery against false information insertion attacks Tarique Haider et al. [6] introduce a popularized fuzzy logic-based concept for power-aware packet routing in WSNs. Ran et al. [7] amends LEACH protocol with fuzzy logic (LEACH-FL), by taking numerous parameters. Salma Mohammedan et al. [8] proposed a totally new methodology, during which fewer operating nodes are selected by using enhanced Ant colony technique. Simulation results depict that the extent of algorithmic complication is depressed and therefore search time is decreased, and the developed algorithm exceeds the other algorithm in condition. Nikravan et al. [9] have approached a fuzzy logic system (FKS) as a selection mechanism for the adjacent node-based path prediction. For every transmission rate and energy, square measures are chosen parameters for selecting the next-hop node in period packet transmission. Experiment results projects that the developed strategy provides enhancement on period transmission and square measure energy economical, whenever it operating on varied real-time atmosphere also planned a new energy economical algorithm called fuzzy dynamic power control algorithm for WSNs [10 –14].

Several proficiencies to optimize energy utilization in WSNs has been developed [15]. These protocols are admitting physical-level design decisions such as transition scaling, voltage exfoliation, etc. to energy cognisant forwarding and energy-aware MAC protocols [16]. Energy-aware routing has begun to receive attention within the recent few years. Whereas ancient routing protocols in wired networks emphasize on increasing point-to-point output and minimizing delay, energy constraints became a significant issue in wireless networks. Several energy-aware routing metrics are suggested [16, 17], so as to reduce energy consumption and enhance the network lifespan. A variety of energy-aware routing protocols is introduced [18 –20]. However, some applications, need the temporal aspects for providing the secure routing based on node availability. So, energy constraints are maintained based on the temporal aspects [27]. And some other works cluster the nodes in based on energy and maintains the path by using graph theory [28]. Sathish et al. [29] provides the dynamic trust modelwith intelligent beta reputation for constructing secure path between wireless nodes. Asanambigai and Ayyasamy [30] proposed impartial clustering technique which helps to maintains less energy consumption and increase the life time of the sensor node.

Van et al. [33] uses the fuzzy based protocol to routing which increase the node lifetime and helps to extend the network lifetime. Song Y [32] discusses the application symmetry algorithm and its techniques in making of optimal allocation of sensor nodes to maintain energy factor. Shi et al. [31] suggested the model to avoid delay and packet loss using fuzzy based approaches. Sathyavathi et al. [34] proposed secure routing for MANET in multiple paths. The routes are formed dynamically based on node energy and trust level.

Although it is seen that there are vast blocks of research being tired in the area of routing problems associated with power consumption in wireless device network victimization of fuzzy logic concept. However, each approach has yielded improvement within the performance victimization through the Fuzzy system got half of the success, but this work proposed protocol will take care of secured packet routing between the sender and receiver using the fuzzy expert system through the mobility prediction approach. Also, it will provide energy balancing among the nodes.

In this research article, a novel intelligent trust-based secured path estimation protocol is aimed that is sub-categorized into three parts: Individual node trust evaluation model trusted and optimal packet route estimation model and effective energy balancing model. In the initial stage of this work have a tendency to establish each node based trust estimation model which comprises the abnormal behavior of the precarious nodes within the open surroundings and therefore the determining imputes of nodes’ tractability. With the help of heuristic-based fuzzy expert (HFE) decision building on the trust estimation properties and expecting at logic rules anticipation, the model will get a trust worth for every node in the networks. The next model considers the trustworthy as the input, a trustworthy routing model is obtained. With the support of the fuzzy expert system, in an effective secured trustworthy packet routing model, which presents a new relied packet routing algorithmic rule is which may throw out the fly-by-night nodes such an authentic packet delivery route is incurred. The third module of this work is going to balance the energy level of the node’s applying altogether with a mobility prediction approach named Intelligent Energy-Aware Multiple restraints Secured Optimal Routing (IEAMSOR) protocol is proposed. The efficiency of this model is compared with traditional protocols using the NS-2.3.5 simulator. The simulation results depict higher detection quantitative relation for untrusted nodes. Moreover, IEAMSOR provides considerable packet delivery magnitude relation; balancing of energy and therefore the network’s output is more efficient when compared with alternative protocols.

The important technical aspect of this work is constituted as follows:

Establish a trust prediction model using a combination of both HFE and the fuzzy logic rules. The whole trust estimation model is a completely hierarchical based structure that consists of base trust, current trust, Gaussian fuzzy membership function, and threshold degree to derive a trust value of each participant in the network.

A suggested trusted packet routing protocol through the assistance of a trust model that is designed previously along with a fuzzy expert system. Parallelly, an IEAMSOR protocol is developed for the purpose of secured and optimal routing purposes. Also, the energy level of all the participant nodes is managed in a well-balanced manner with the help of a mobility estimation approach.

Developed a simulation model through NS 2.3.5 in order to validate the performance of this proposed protocol in malicious node analysis and stability against the untrusted node attacks.

The remainder of the paper is integrated as follows. In section 2, convey the trust estimation model in a concise manner along with heuristic-based fuzzy expert system concepts, and constitute a trusted routing model and energy balancing among the nodes. The experimental results, as well as the various performance measures, are depicted in section 3. Finally, section 4 describes the conclusion and future extension of this research work.

2 Methodologies and system models

2.1 Trust construction model

Trust manifests the assumption or assurance or anticipations on the honestness, unity, power, handiness, and quality of service of objective node’s succeeding activeness. It additionally shines the reciprocal relationships wherever an afforded node acts in an exceedingly trustful manner and holds authentic communications alone with nodes that are extremely believed by the committed node.

Trust measure conveys the range that one particular node requires other nodes to extend supply bound services. At present trustworthy management protocols target a way to appraise and develop an exact trustworthy measure, and the way to approach the ends up in trustworthy applications. Estimating nodes must quantify all relevant data concerning associate evaluated node, together with the observations node’s behaviors, interacting records, view from alternative nodes, and so on. It employs an acceptable model to measure the credibility of the judged node. The trustworthy measurements are very useful in routing call that acts as a peculiar trustworthy application.

2.1.1 Categorization of specific kinds of Trust

This Heuristic-based trust model consists of three different types of trust values, such as basic trust value, current trust value and route trust value also called path trust value. In this new model, we first compute the basic trust using a direct discussion with the neighbors. In each node, an intelligent agent is deployed in order to compute the basic trust and to maintain history about the neighbors. This basic trust is updated dynamically based on the communication using the metrics namely energy consumed, delay values, number of packets dropped by the node, the capacity of the node and cooperation of the node with its neighbors. The updated trust values are known as current trust. There are two types of trusts namely basic and current trust are represented by BT_ij and current trust denoted by CT_ij. The node trust is computed by using basic trust and current trust and is denoted by NTV_ij .

2.1.1.1. Node’s Basic Trust Estimation. Node trust is estimated by the node’s physical neighbors based on basic interaction information at the end of each time interval, which is calculated by two decision factors. These factors are basic trust denoted by BT_ij and current trust denoted by CT_ij. The basic trust value BT_ij denotes node v_j’s trust level from the evaluating node v_i’s point of view, which can be calculated by ${NTV}_{ij} (t) = α {BT}_{ij} (t) + β {CT}_{ij} (t) t_{1} ⩽ t ⩽ t_{2}$ (1)

The weights α and β (α, β> = 0, α> β, and α + β= 1) are assigned to BT_ij and CT_ij. Now the basic trust is computed using the relation represented by SE_m(i,j) ${BT}_{ij}^{t_{k}} (t) = \frac{\sum_{m = 1}^{N_{t_{k}}} {SE}_{m} (i, j)}{N_{t_{k}}}$ (2)

Where

BT_ij –Basic trust

SE_m (i, j) –trust between i and j node

2.1.1.2. Node’s Current Trust Estimation. The current Trust value is estimated in this model is the trust value of the node in the time interval between t and t + 1. This proposed trust model from this research work computes the node’s entire trust value based on the fuzzy expert system approach. In this article, the term current trust (CT) represents the node’s current trust value. Creditability is another factor established by trust using a threshold. If a node has a trust value greater than the threshold, then its creditability is high else if it is near the threshold and goes up and down dynamically, it is called medium. Otherwise, it is called low. ${CT}_{irj} (t) = {CC}_{ir} (t) \times {BT}_{rj} (t) t_{1} ⩽ t ⩽ NOW$ (3)

In this work, current trust is computed using the following mathematical representation is

$\begin{matrix} {CC}_{ir} (t) = {BT}_{i 1} (t) \times {BT}_{i 2} (t) \times {BT}_{i 3} (t) \\ \times . . . . . . . . \times {BT}_{(r - 1) r} (t) t_{1} ⩽ t ⩽ NOW \end{matrix}$ (4)

If n nodes are present in the communication, we have current trust values:

CT_rP1j (t) , CT_iP2j (t) , . . . . . . , CT_iPnj (t). Using these n values, CT(t) is computed using the form ${CT}_{ij} (t) = \sum_{k = 1}^{n} W_{P_{k}} \times {CT}_{{iP}_{kj}} (t)$ (5)

This model estimates the trust value of the node i on node j in time interval t + 1 is represented as (T_ij (t + 1)) is derived with the help of both basic trust of i on j at time t (BT_ij (t)) and current trust on j to i by few other nodes at the time of t as (CT_ij (t)) as shown in the equation as follows

$\begin{matrix} T^{i} (t + 1) = α \times {BT}_{j}^{i} (t) + (1 - α) \times {CT}_{j}^{i} (t), \\ 0 ⩽ α ⩽ 1, t_{1} ⩽ t ⩽ NOW \end{matrix}$ (6)

2.2 Route trust estimation

Whenever a sender node detects a route to the receiver node with the assistance of furtherance nodes, the trustworthy value of the route (path trust) ought to be calculated consistent with the trust values of the nodes existing on that specific route. The concatenation multiplication of trust does not increase the trust value, route trust shouldn’t be quite the trust measures of the arbitrate nodes. Therefore, at a time interval t, the trust of a route P represented by P_T (t) is up to the continuing production of node trust values within the route, that is $\begin{matrix} P_{T} (t) = \prod \\ ({{NTV}_{ij} (t) | n_{i}, n_{j} \in P and n_{i} \to n_{j} and n_{j} \neq N_{d}}) \end{matrix}$ (7)

Here the t^th interval attenuation range function is compared with recent t + 1 interval in the trust estimation is specified as the time decomposition mathematical function. ${df}_{t} = ρ^{n - t}, 0 < ρ < 1, 1 ⩽ t ⩽ 1$ (8)

The basis constant ρ is constituted by the attenuation range factor. A minimum ρ induces a bigger attenuation range of df_t, and the other way around. At last, node v_i compute a base trust vale for node v_j in keeping with their account of interaction estimations through the following mathematical form ${BT}_{ij} = \frac{\sum_{t = 1}^{n} {df}_{t} \times C_{ij}^{tk} \times (1 - e^{- (N_{t_{k}} / σ)})}{\sum_{t = 1}^{n} {df}_{t} \times (1 - e^{- (N_{t_{k}} / σ)})}$ (9)

Regulative issue σ is employed to measure the effects of a range of interactions on the basic trust calculation, and therefore the interaction component is referred by (1 - e ^-(N_tk/σ)). The solidify measure of σ is aligned based on the atmosphere and features of the practical application. The interaction factor encompasses a negative exponential increment to the number of interactions in an established duration. This issue is employed to stress the grandness of the group action range. The basic trustworthy value is the weighted average of all interaction ratings at dissimilar interaction ranges.

2.3 Energy model

Energy used by the nodes of the sensor has been obtained and this is used for the threshold computation. This threshold value plays a vital role in the election of the cluster head (CH) node and optimization of the cluster. The energy used by the node is mainly due to the computation and transmission of the data. The communication energy of the sensor node is high while comparing to the computational energy. In this work, the energy model of the sensor network has been designed with communication energy alone. The threshold computational technique proposed in this work is as follows: Let E be the present energy available in the node. The transmission and the receiving energy of the node during communication in the network are E_t and E_r respectively. Then, the total energy consumed (EC) by the node used for communication is given by $E_{c} = E_{r} + E_{t}$ (10)

Each node has a weight factor (K) that is used for the computation of the threshold value. This weight factor is obtained by combining the parameters, x bits of data transmitted of data over a distance (d), the energy (ξ) spent for transmitting the x bits per unit distance. Then, the weight factor of the i^th node is given by $K_{i} = d_{i} * x * ξ$ (11) where d_i is the distance of i^th node to the base station.

Based on the weight factor and energy consumption, the threshold value (Th) at any particular time Δt as given in Equation 12 and also shown in Fig. 1. $Th = \frac{E_{n} - max {K_{i} * E_{c} * t} * n}{n}$ (12)

Fig. 1

Threshold Computation.

where n is the cluster size,

E_n is the total energy available in the network,

K_i the weight factor of the node,

E_c is the energy consumption at time t.

For a particular node, during the election process, the ratio (R_i) between the distance from the Base Station (d_BS) and (E_i) the energy of the i^th node in the network is computed as given in Equation 13

$\begin{matrix} R_{i} = {[\frac{d_{i}}{E_{i}}]}_{i = 0}^{n} \\ (excluding the CH and broad cast channel [BCH]) \end{matrix}$ (13) From Equation 13, the minimum value node id is obtained. Then, their individual threshold valves (Th_i) are computed as at time t is given Equation 14. ${Th}_{i} = E_{i} - (K_{i} * E_{c} * t)$ (14)

From Equation 14 the individual Th_i of i^th is compared with the overall network threshold value to select a cluster head and backup cluster head. In this research work, two different values of threshold Th₀ and Th₁ were used for initial and subsequent cluster formation respectively. The value Th₀ is computed based on the distance of the node from the base station and assuming the nodes have the same energy level. ${Th}_{0} = min {d_{i}} i = iton$ (15) where d_i is the distance of i^th node from the base station.

During the initial stage of clustering, the node with the distance that is less than or equal to the threshold value (Th₀) may be elected as a CH. Then, the subsequent stages the CH is selected by comparing their metric value with the threshold value (Th₁) obtained using Equation 12.

2.4 Fuzzy expert system

It is a set of membership functions and the rules that are used to reason about given data [21, 22]. It uses fuzzy logic concepts instead of Boolean logic. It is not like symbolic reasoning engines, fuzzy expert systems are purely numerical oriented processing. The accuracy and success of a fuzzy expert system completely depend upon the belief of the domain experts based on the issues regarding the research work. This system incorporated the following functional units such as the knowledge base, Fuzzification process, Fuzzy logic inference engine, and the Defuzzification process. The pictorial representation of the entire components is shown in Fig. 2.

Fig. 2

Diagram representation of Fuzzy expert system.

2.4.1 Fuzzification

The process of transforming crisp values into grades of membership for linguistic terms of fuzzy sets is known as fuzzification. It can be performed by approaching many of the real-world quantities in the form of a crisp set. In general, the crisp set of existing data may be in the form of an uncertain manner, because of imprecision, ambiguity, or vagueness. In order to resolve these problems, it can be transformed into fuzzy by various types of membership functions. Later this approach is called fuzzifiers. There are four different types of fuzzifiers in general, which are most popularly used for the fuzzification process, such as Gaussian fuzzifiers, Trapezoidal fuzzifiers, Triangular fuzzifiers and Singleton fuzzifiers [23, 24].

In this research work, the proposed model incorporates Gaussian fuzzifiers for estimating membership values of the truest measures of specific nodes. According to the Gaussian membership function, the membership functions of fuzzy input sets are depended upon two various arguments, standard deviation σ and mean c. The mathematical form of the Gaussian fuzzifier is shown below $μ_{Trust - value} (X) = e^{(\frac{- {(x - c)}^{2}}{2 σ^{2}})}$ (16)

The initial step in the development of a fuzzy logic-based expert system is to generate a fuzzy set for the given node based trust parameters. This process can be performed by equation 16. Based on the knowledge of domain experts, input parameters (Low, Low-Medium, Medium, and High), as well as output parameters (Low, Low-Medium, Medium, and High), are selected. The range of fuzzy value for each linguistic variable of a trust-based parameter is shown in Tables 1. The fuzzification process begins with the transubstantiation of the given node based trust parameters using the functions that are represented in equation 16.

Table 1

The range of fuzzy values for each input trust parameter Basic Trust (BT), Current Trust (CT) and Path-Trust (P_T)

Linguistic Variables	Fuzzy Values	Symbols
Low	0.0≤z≤0.4	LOW
Low-Medium	0.3≤z≤0.6	LM
Medium	0.5≤z≤0.8	MED
High	0.7≤z≤1.0	HGH

During this process, linguistic variables are evaluated using the Gaussian membership function and accompanied by a degree of membership ranging from 0 to 1. The proposed model divided the entire parameters into four categories according to its degree of membership values. While contriving a fuzzy inference system, it is very easy to empathize that membership functions are related to linguistic term sets, which commonly seem in the antecedent part of the rule otherwise consequent part of the rule. This mathematical formula is identified based on the recommendation given by the domain experts and various data analyses performed in the previous section in a heuristic manner. In a similar fashion, this system has divided current trust (CT) into four categories. Both Basic and Current trust of node’s related fuzzy membership representation is shown in Figs. 3 & 4 respectively.

Fig. 3

Fuzzy membership function representation of the Node’s Basic Trust (BT).

Fig. 4

Fuzzy membership function representation of the Node’s Current Trust (CT).

After the fuzzification process, the proposed model going to generate a set of fuzzy rules. The fuzzy rules for this user model were developed with the assistance of domain experts. The knowledge-base of fuzzy expert system based dynamic trusted routing (FESDTR) as much as fuzzy rules, generated with the help of combination theory. Here a set of valid rules is selected by the domain experts. These rules are experimentally tested and verified by the various statistical measures via Statistical Package for the Social Sciences (SPSS) and also considered fuzzy partition methods for deriving linguistic variables based on the node’s trust parameters.

The process of inferring a conclusion from the existing data is known as inference. In other words, computing output values from the given input values applying the knowledge are called inference. Fuzzy inference is the process of mapping from a given input into an output using the theory of fuzzy sets [24 –26]. The central tendency of the inference engine is to make the decisions using the rules which exist in the rule base. The fuzzy inference engine from this user model uses the rules present in the knowledge-base and generates a conclusion based on the rules.

The rules use the input membership values as weighting factors to predict their influence on the fuzzy output sets of the final output decision. The degrees of truths (R) of the fuzzy rules are determined for each rule by evaluating the non-zero minimum values using the logical AND operator. Only the rules that get strength greater than 0, will be firing the output. In this research work approach Root Sum Square (RSS) employed an inference engine to estimate the firing values of the rules. The RSS is represented by equation 17. $RSS = \sqrt{\sum R^{2}} = \sqrt{(R_{1}^{2} + R_{2}^{2} + R_{3}^{2} + . . . + R_{n}^{2})}$ (17)

$R_{1}^{2} + R_{2}^{2} + R_{3}^{2} + . . . + R_{n}^{2}$ where are strength values of different rules which share the same conclusion. This method aggregates the effects of all the applicable fuzzy rules, scales the functions at their respective magnitudes, and calculates the “fuzzy” centroid of the composite area. RSS method is more complicated mathematically than other methods [30 –32], but it is selected for this research work since it gives the acceptable weighted influence of all firing rules. Experimental of the rule base evaluation for the nodes trust vales are shown. The Root Sum Square-based inference drawing is found to be the most suitable approach to infer data from the fuzzy rules developed. According to the rules discussed above, the proposed model consists of 16 distinct fuzzy rules from the rule base of the fuzzy expert system, which are represented below:

R₁: IF BT = Low AND RT = Low THEN Node Trust = Low

R₂: IF BT = Low Medium AND RT = Low THEN Node Trust = Low

R₃: IF BT = Medium AND RT = Low THEN Node Trust = Low Medium

R₄: IF BT = High AND RT = Low THEN Node Trust = Medium

R₅: IF BT = Low AND RT = Low Medium THEN Node Trust = Low

R₆: IF BT = Low Medium AND RT = Low Medium THEN Node Trust = Low Medium

R₇: IF BT = Medium AND RT = Low Medium THEN Node Trust = Low Medium

R₈: IF BT = High AND RT = Low Medium THEN Node Trust = Medium

R₉: IF BT = Low AND RT = Medium THEN Node Trust = Low

R₁₀: IF BT = Low Medium AND RT = Medium THEN Node Trust = Low Medium

R₁₁: IF BT = Medium AND RT = Medium THEN Node Trust = Medium

R₁₂: IF BT = High AND RT = Medium THEN Node Trust = Medium

R₁₃: IF BT = Low AND RT = High THEN Node Trust = Low

R₁₄: IF BT = Low Medium AND RT = High THEN Node Trust = Low Medium

R₁₅: IF BT = Medium AND RT = High THEN Node Trust = Medium

R₁₆: IF BT = High AND RT = High THEN Node Trust = High

The proposed model considers both path-trust (P_T) and a number of un-trusted nodes present in the path between source to destination to calculate each path trust value and finally recommend the optimal trusted path between the desired source to destination using a fuzzy expert system approach.

The range of fuzzy value for each linguistic variable of Path-Trust based parameter and a number of un-trusted nodes related to fuzzy value for each linguistic variable are shown in Tables 1 and 2 respectively. Moreover, the fuzzy rules are derived with the help of both path trust (P_T) and the number of untrusted nodes (UTN). Also, their fuzzy linguistic variable based fuzzy membership degree values are shown in Figs. 5 and 6 respectively.

Table 2

The range of fuzzy values for each input trust parameter for Number of un-trusted nodes

Linguistic Variables	Fuzzy Values	Symbols
Low	0.8≤z≤0.9	LOW
Low-Medium	0.7≤z≤0.6	LM
Medium	0.5≤z≤0.4	MED
High	0.3≤z≤0.1	HGH

Fig. 5

Fuzzy membership function representation for Path-Trust (P_T).

Fig. 6

Fuzzy membership function representation for Number of Untrusted Nodes (UTN).

2.5 Mobility prediction mechanism

In this proposed model, the stability of a data transmission path is based on the dependability or accessibility of every link of the particular route due to the dynamism in a network topology. Also presume a free space extension model [15], here the received signal strength exclusively reckons on the distance to the sender node. Hence, applying the motion parameters of two adjacent nodes like speed, motion direction, and communication propagation range are estimated well in advance, the time duration of these pairs of nodes will be calculated based on their connection period. Let two adjacent nodes i and j are within the transmission limit T_L among them. Assume (X_i, Y_i) and (X_j, Y_j) be the co-ordinate points of wandering node j. In addition, assume (V_i, Θ_i) becomes the speed and the movement direction of the sender node i, similarly (V_j, Θ_j) be the speed and the motion direction of the receiver node j. Hence, the amount of two mobile nodes will remain connected and is calculated by: $RFT = \frac{- (ab + cd) + \sqrt{(a^{2} + c^{2}) T_{L}^{2} - (ad - bc)^{2}}}{a^{2} + c^{2}}$ (18)

where $\begin{matrix} a = v_{i} cos θ_{i} - v_{j} cos θ_{j}, \\ b = x_{i} - x_{j}, \\ c = v_{i} sin θ_{i} - v_{j} sin θ_{j}, and \\ d = y_{i} - y_{j} \end{matrix}$

Suppose that when v_i = v_j and Θ_i = Θ_j, then RFT becomes infinity. The estimated value is the Route Failure Time (RFT) among the two adjacent nodes. The above-mentioned equation form only identifies the lifetime of the link established between the two adjacent neighbor nodes. Suppose that any one of the nodes connected on a link abruptly alters its speed, moving direction or both, then the RFT consociated with that link will also be needed to be modified. Hence, the dynamism in WSNs is unable to cover by the above equation 18. This problem is examined in the following two scenarios:

Scenario 1: Suppose both i and j active nodes are anticipated to increase or decrease their speed, moving direction or both during their communication in mobility.

Scenario 2: If anyone of the nodes or both decaying their energy level due to their malicious behavior.

In the above two scenarios, because of the prominent dynamic property of the node mobility, the estimation of RFT regarding that particular link will be a tedious task. It will pretend the availability of the link between the two adjacent nodes. Moreover, despite that the fact RFT is eminent, suppose any one of the nodes or both the nodes communication links are insufficient energy levels due to their malicious behaviors may lead to losing the connection between them. This situation also impresses the availability of the path. Therefore the estimated RFT of that link will mislead the entire model.

Another point of view, the nodes on a particular link might have a sensible energy level and that they might forward a huge amount of packets. Suppose any one of the nodes or both the nodes and their link may be disconnected to fulminant alteration in its speed, moving direction or both during the node mobility, the packet delivery ratio (PDR) on that particular link is drastically reduced. It pretends PDR on the selected path. Therefore, the RFT estimation mentioned in equation 18 is required modification due to the dynamic mobility in WSNs.

In order to optimize the energy level of each node in the WSNs, the proposed protocol needed a modified and novel mobility prediction formula to estimate the availability of RFT is given in equation 19. The time duration required to lose the entire energy in a particular node is referred to as Life Time Energy (LTE) of the node. In general, any node will spend its energy throughout any one of the following modes during their mobility such as Ideal mode, Pre-processing mode, and Transfer mode. Therefore, the aggregation of all these three mode energy loss will provide the exact life period of a particular node shown in equation 20. $RFT = \frac{- (ab + cd) + \sqrt{(a^{2} + c^{2}) T_{L}^{2} - (ad - bc)^{2}}}{a^{2} + c^{2}} + LTE (n_{i})$ (19) $LTE (n_{i}) = [(α \times μ (e_{i})) + (β \times μ (m_{i}))] + \sum_{i = 1}^{tn} (I_{m} (i) + {Pp}_{m} (i) + T_{m} (i))$ (20)

where

α -Weight parameter based on the node behavior

β –Weight parameter based on the Neighbour suggestion $μ (e_{i} = δ e_{i}, 1 ⩽ i ⩽ n δ - control parameter)$ $μ (m_{i}) = β (NP \times λ)$ $λ = \frac{\sum_{i = 1}^{n} μ (e_{i})}{n}$

Therefore in the above two scenarios, prediction of RFT of each path between pair of node mention in equation 19 is fine-tuned by a desirable variable known as Mobility Adjustment Parameter (MAP) during node’s active mobility situation. Hence, the RFT value will be calculated by contributing the value of MAP additionally along with RFT is named as EstimatedRFT (ERFT). The various parameters used for fine-tuning the MAP are listed in Table 3. The EstimatedRFT value is calculated through the following equation 21 as shown.

Table 3

Various parameters involved in the fine-tuning process of MAP

Node ID	Sender Node Trust Class	Receiver Node Trust Class	Mobility Node (s)	No. of Neighbour nodes
N₁	MED	MED	SENDER/RECEIVER	3
N₂	MED	MED	SENDER/RECEIVER	2
N₃	MED	MED	SENDER AND RECEIVER	2
N₄	MED	HGH	SENDER AND RECEIVER	4
N₅	MED	LOW	SENDER AND RECEIVER	4
N₆	HGH	MED	SENDER/RECEIVER	4
N₇	MED	MED	SENDER/RECEIVER	5
N₈	LOW	HGH	SENDER/RECEIVER	2
N₉	MED	MED	SENDER AND RECEIVER	3
N₁₀	HGH	MED	SENDER AND RECEIVER	3

$EstimatedRFT = RFT + MAP$ (21)

where $MAP = \frac{T_{MT}}{T_{ST} + T_{MT}}$ (22)

Proposed network model: The proposed network model in WSNs is represented by G = V, E where V is the set of interconnected vertices (nodes) and E is the set of full-duplex directed wireless communication edges (links). The network model considers the existence of multiple paths between any two nodes where each link on each path considers the Quality of Service (QoS) metrics namely Delay (D), HopCount (HC), and PathOptimality (PO). This model also considers the EL of each node (V_i) on each path in the network topology based on their trustworthiness (Low, Low-Medium, Medium, and High) using fuzzy rules.

The proposed model considers that the energy level (EL) of every node and its corresponding path must meet a certain amount of energy level. Here the every individual node energy level is based on its residual battery backup, which is gathered and added together for the purpose of calculating the energy level of every transmission path. Hence, the energy level of each routing path is represented by the following form $P_{EL} = \frac{\sum_{i = 1}^{n} EnergyLevel (V_{i})}{n}$

It also incorporates the additional approach called EstimatedRFT parameter in order to select the best optimal path among the n-number of paths between sender to receiver node is represented as P₁, P₂, P₃,... and P_n. Here the optimal path P_k is one that satisfies the above mentioned QoS parameters minimum threshold level.

2.6 Intelligent energy aware multiple restraints secured optimal routing (ieamsor) protocol with dynamic mobility estimation

The route discovery and route maintenance process of this proposed protocol is shown below. In the process of route discovery, the best optimal is chosen based on the various QoS restraints and energy criteria through the concept of mobility prediction. Hence,

the route with higher battery life and maximum predictedRFT will be recognized as a solid optimal path between the sender and receiver.

Route Discovery: In this proposed novel approach, the sender node flooding a Route Establish Request (RERQ) packet, which consists: Sender_address, Receiver_address, Packet_Type, Route_Request_ID, average delay (Dc), Energy level (EL), and Hop Count (HC). Suppose this packet is received by intermediate nodes then it will simply forwarding it into each out link when the link energy level greater than the average EL. This approach will reduce the unnecessary forwarding of RERQs during the pathfinding process, also overcomes the control overhead.

Suppose the receiver node receives a duplicate RERQ then it will be tossed out. On the other hand, the receiver node generates a Route Establish Request-Reply (RERR) which consists of Sender_address, Receiver_address, Packet_Type, Route_Request_ID, PredictedRTL, EL, HC and Mobiliy_Information. Thereafter the receiver node broadcast RERR to the sender node. The proposed protocol manages the neighbor node table which consists of each intermediate node related complete details such as Route_ID, EL, HC, Sender_address, and Receiver_address. These details are updated when any changes occurred in the network topology by the sink node. Therefore, the unwanted packet flooding is reduced drastically also network overhead is completely cut down. Whenever an intermediate node receives a route request reply from the other node, it will estimate the RFT for the corresponding link using equation 19.

The sender node also maintains its own table which consists of the following parameters such as Delay (D), Hope Count (HC), Energy Level (EL), and EstimatedRFT. These parameters are updated whenever it receives each RERR from the receiver node. Updated parameters are compared with the threshold parameters such as average delay (Dc), Hop Count (HC), etc. Suppose the comparison is successful then it will select the best optimal path among the n-number of existing paths between the sender and receiver. The selected path satisfies the basic QoS restraints to the receiver node. The path discovery process for the projected protocol is depicted as follows:

Procedure for Sender (S)

If sender S has no route to Receiver R

Initialize the Optimal Path QoS restraints

Build and flood RERQ

Perform path return management algorithm

Accomplish path maintenance

End if

Path Postulation Management Algorithm

If current node is intermediate node IMN (Intermediate Node)

If the obtained RERQ not duplicate

If ((D< = D_c) && (HC< = HC_c) && (PO_ij> = APO_c))

Forward RERR message

End if

If the current node is receiver node R

If the received RERQ is original

Run the path reply maintenance algorithm

End if

Path Response Management Procedure

If current node is receiver R

If obtained RERQ is original

Initialize all Optimal path parameters into 0 (HC, RFT, etc)

Collect its speed and moving direction through the positioning system

Generate RERR which contains updated EL, RFT, etc

Forward RERR in the direction of Sender S

End if

If the current node is an intermediate node IMN

If accepted RERR is original and PO_ij> = APO_c

Update all Optimal Path QoS constraints

Collect its speed and moving direction through the positioning system

Estimate updated RFT

Generate RERR which contains updated EL, RFT, etc to Sender S

Forward the same to S

End if

If the node is Sender S

Accept the RERRs

Accumulate the routes to Receiver R

If the path table is not null then sort based on their optimal value

For each route P_i

If HC_i < = HC_c and RFT_i > 0 then select the Path P_i , include it in the route selection table

else remove the path P_i

End if

End for

For every path P_i in the route selection table

If RFT (P_i)> = RFT (P_1, P_2, ... .. P_n)

Choose P_i for further packet transmission

End if

End for

End if

Route Maintenance: Because of the dynamism in network topology and restriction in network resources, the estimated and selected optimal path much becomes voided. Suppose the current data link is disconnected then the intermediate node transmits the Route Establish Reconstruction Packet (RERC) to the sender. Thereafter, the sender node repeats the execution of the route discovery process. Suppose the sender node receives REPP and RERC simultaneously then it carries the RERC.

Route maintenance procedure by an intermediate node

If the link is disconnected with its adjacent node

Build and forward RERC to Sender S

End if

If the RERC is got from its adjacent node

Forward the RERC to Sender S

End if

Route maintenance procedure by Sender node

If the RERC is accepted from any IMN (Intermediate node)

If a backup route is available

Transmit the packets through a backup path

else flooding the RERQ

End if

If RERC and RERR is received simultaneously from any intermediate node IMN

Flooding New RERQ

End if

Consider the diagram representation of path establishment among the 10 number of nodes with their energy level shown in Fig. 7. In this example, the path P₃ and P₁₈ are chosen as optimal paths between the sender node N₁ and the receiver node N₁₀. Because of these two paths (P₃ and P₁₈) satisfies the entire threshold QoS restraints. Table 4 provides the details of the latest RFT values of these paths P₃ and P₁₈ respectively based on equation 19 along with their energy level (EL) prior to its mobility. The recentRFT of the paths P₃ and P₁₈ are estimated at 2.5090 and 2.5096 respectively also met the criteria RFTs are > 0.

Fig. 7

Multiple QoS path between Sender and receiver of the network topology.

Table 4

Updated recentRFT for P₃ and P₁₈ without mobility

Path P ₃ (1, 4, 7, 10)
Link	X _i	Y _i	X _j	Y _j	Θ _i	Θ _j	V _i	V _j	r	EL _(i)	EL _(j)	RecentRFT
1–4	120	120	170	170	100	90	11	65	160	93	90	2.8843
4–7	170	170	270	170	65	50	15	35	140	90	83	2.5090
7–10	270	170	320	120	75	50	30	20	100	83	84	5.9896
RecentRFT of P ₃ (1, 4, 7, 10)												2.5090
Path P ₁₈ (1, 4, 7, 9, 10)
Link	X _i	Y _i	X _j	Y _j	Θ _i	Θ _j	V _i	V _j	r	EL _(i)	EL _(j)	RecentRFT
1–4	120	120	170	170	100	90	11	65	160	93	90	2.8843
4–7	170	170	270	170	65	50	45	9	140	90	83	2.5094
7–9	270	170	270	70	90	60	9	45	155	83	80	3.6559
9-10	270	70	320	120	90	50	65	20	150	80	84	2.9214
RecentRFT of P ₁₈ (1, 4, 7, 9, 10)												2.5094

Table 5 depicts the recent RFT for the paths P3 and P18 based on the mathematical notation 21 after their mobility along with their respective modified EL. In order to achieve power consumption during mobility, the proposed protocol approaches equation 21 to balance the energy of the nodes existing among the optimal paths based on their trust class. Here this novel approach assigns maximum energy weight to a minimum (α, β) based up their trust class (like Low, Low-Medium, Medium, and High) therefore the balancing of energy between the highly trusted nodes is increased at the same time the malicious nodes EL is decreased. Hence, the energy balancing is achieved by this proposed novel protocol through a fuzzy rule-based trust prediction approach is shown in Table 6. Also, the calculation of estimated RTT of each link on P₃ and P₁₈ during mobility with a MAP of those paths based on the mathematical notation 21 is represented in Table 6.

Table 5

Updated recentRFT for P₃ and P₁₈ after mobility

Path P ₃ (1, 4, 7, 10)
Link	X _i	Y _i	X _j	Y _j	Θ _i	Θ _j	V _i	V _j	r	EL _(i)	EL _(j)	RecentRFT
1–4	140	110	170	170	100	90	25	65	160	83	87	3.7545
4–7	160	200	270	240	120	50	15	20	140	87	80	2.6847
7–10	270	170	320	120	75	50	25	20	100	80	79	6.9635
RecentRFT of P ₃ (1, 4, 7, 10)												2.6847
Path P ₁₈ (1, 4, 7, 9, 10)
Link	X _i	Y _i	X _j	Y _j	Θ _i	Θ _j	V _i	V _j	r	EL _(i)	EL _(j)	RecentRFT
1–4	160	200	270	240	100	90	27	65	140	78	87	3.1236
4–7	170	170	270	170	65	50	45	25	140	87	78	6.8965
7–9	290	200	240	80	90	60	30	45	155	78	78	7.9729
9-10	300	90	250	110	80	50	80	40	150	78	74	3.498
RecentRFT of P ₁₈ (1, 4, 7, 9, 10)												3.1236

Table 6

EstimatedRFT for P₃ and P₁₈ after mobility using Equation 21

Path	Sender	Receiver	Mobile nodes of the link	Nearer to the neighbor node	Departs from a neighbor node	MAF	Estimated RFT
1–4	MED	HGH	SENDER	YES	NO	0.592	4.3465
4–7	MED	MED	RECEIVER	YES	NO	3.115	5.7997
7–10	MED	MED	SENDER	YES	NO	1.82	8.7845
EstimatedRFT of P ₃ (1, 4, 7, 10)							4.3465
1–4	MED	HGH	SENDER	YES	NO	1.496	4.6196
4–7	MED	MED	RECEIVER	YES	NO	2.537	9.4335
7–9	MED	MED	SENDER	YES	NO	1.674	9.6469
9-10	MED	MED	SENDER AND RECEIVER	YES	NO	4.795	8.2930
EstimatedRFT of P ₁₈ (1, 4, 7, 9, 10)							4.6196

3 Result and discussion

The various systems of measures used for determining the various protocols are packet transmission, Transmission overhead, Energy level, Energy consumption, and Network Lifetime analysis. These measures were equated with additional metrics such

as a number of moving nodes and node mobility speed. The proposed protocol is compared with TBP, MQRPMP, PMQRMP, and EMQRPDM [15] is shown in Figs. 8–12. Figure 8 represents the packet transmission rate versus node’s moving speed. Packet transmission overhead along with a number of active nodes is depicted in Fig. 9. Mobility speed versus energy level comparison, Network lifetime and Energy consumption with a number of active nodes are presented in Figs. 10, 11 and 12 respectively.

The comparison of packet transmission rate versus node’s mobility is shown in Fig. 8. While the node’s mobility speed is 2 Meter/Sec then the packet transmission rate of the proposed protocol achieved the value 0.99 which is greater than the other existing protocols such as TBP, MQRPMP, PMQRMP, and EMQRPDM [15]. whenever the node’s moving speed increased more than 2 Meter/Sec, the performance of existing protocols is reduced extremely down. Nevertheless, the proposed protocol keeps on increasing and remains greater than the existing protocols. Finally, it reaches 0.87 if the node’s moving speed is 10 Meter/Sec, because of the computation of EstimatedRFT along with fuzzy based node trust classification approach. Here the comparison over the cost of control overhead with respect to a number of active nodes is identified during transmission of existing protocols with the proposed novel protocol.

Fig. 8

Packet transmission rate versus node’s moving speed.

While raising the number of active nodes in the topology, the transmission cost of control packets also increases (i.e., both are directly proportional). When compared with existing protocols the proposed one is very less overhead than others is represented in Fig. 9.

Fig. 9

Packet transmission overhead along with a number of active nodes.

The main objective of this proposed protocol is to provide optimal energy consumption during the node’s mobility, which is achieved using energy balanced approach through adding energy weight values to the nodes which are highly trusted and decreasing their energy weight while their trust is low-medium or low. In this work, the proposed protocol consumes considerable power even the nodes are in moving when compared with existing protocols is visually represented in Fig. 10.

Fig. 10

Mobility speed versus energy level comparison.

Even the number of active nodes increases in the network does not affect the EstimatedRFT of the optimal path between the sender and receiver due to the energy balancing approach incorporated in this approach also shown in Fig. 11. Due to this reason, the network lifetime of this novel protocol is much higher than the other traditional protocols mentioned in the earlier section. The enormous variation existing in this measure when compared with existing protocols with the proposed one is depicted in Fig. 12. Moreover, this model underperforms whenever the nodes are suddenly get failed.

Fig. 11

Mobility speed versus energy level comparison.

Fig. 12

Energy consumption with a number of active nodes.

4 Conclusion and future work

The proposed work discussed the novel protocol IEAMSOR with various QoS restraints between sender and receiver nodes during their communication. The main objective of this protocol is that conceives a reasonable amount of energy during the packet transmission between the sender and receiver. It also incorporated the Node’s Mobility estimation approach in order to estimate the optimal energy route between the communication nodes. The novel IEAMSOR furnishes an immediate response while any changes that happened in the network decreases the unwanted usage of resources and raises substantial betterment in packet transmission rate and thus drastically cut down the network overhead as well as control overhead to rebuild the optimal route establishment.

It also comprises the fuzzy rule-based node trust estimation and optimal path prediction provides secured packet transmission between the sender and receiver hence the energy balance is achieved. Therefore, this proposed protocol raises a multiple path packet routing along with secured and intelligent overhead distribution among the nodes. Further, this protocol will be merged with other hybrid packet routing protocols from WSNs in order to enhance the Packet Delivery Ratio along with additional QoS restraints through a hybrid combination of soft computing techniques. Also, to predict the abrupt failures of the nodes in the networks.

Footnotes

Acknowledgments

One of the authors K.Selvakumar is thankful to the UGC, New Delhi India, for funding through UGC-BSR fellowship to carry out this research work.

References

Elrahim

A.G.A.

, Elsayed

H.A.

, El Ramly

and Ibrahim

M.M.

, An Energy-Aware WSN Geographic Routing Protocol, Universal Journal of Computer Science and Engineering Technology 1(2) (2010), 105–111.

Selvakumar

, Sairamesh

and Kannan

, An intelligent energy aware secured algorithm for routing in wireless sensor networks, Wireless Personal Communications 96(3) (2017), 4781–4798.

Swapna Kumar

, Nanda Kumar

, Sheeba

V.S.

and Kashwan

K.R.

, Cluster-Based Routing Algorithm Using Dual Staged Fuzzy Logic in Wireless Sensor Networks, Journal of Information & Computational Science 9(5) (2012), 1281–1297.

Ayyasamy

, Archana

, Design and analysis of QoS for different routing protocol in mobile ad hoc networks. In Next-generation networks, pp. 247–253. Springer, Singapore, 2018.

Sun

C.I.

, Lee

H.Y.

and Cho

T.H.

, A path selection method for improving the detection power of statistical filtering in sensor networks, Journal of Information Science and Engineering 25 (2009), 1163–1175.

Haider

and Yusuf

, A fuzzy approach to energy-optimized routing for wireless sensor networks, The International Arab Journal of Information Technology 6(2) (2009), 179–188.

Ran

, Zhang

and Gong

, Improving on LEACH protocol of wireless sensor networks using fuzzy logic, Journal of Information & Computational Science 7(3) (2010), 767–775.

Begum

, Tara

and Sultana

, Energy-efficient target coverage in wireless sensor networks based on modified Ant Colony Algorithm, International Journal of Ad Hoc, Sensor & Ubiquitous Computing 1(4) (2010), 29–36.

Nikravan

, Jameii

S.M.

and Haghi Kashani

, An intelligent energy-efficient QoS-routing scheme for WSN, International Journal of Advanced Engineering Sciences and Technologies 8(1) (2011), 121–124.

10.

Akkaya

and Younis

, Energy-aware, delay-constrained routing in WSNs through Genetic Algorithm, International Journal of Communication Systems 17(6) (2004), 663–687.

11.

Sethukkarasi

, Ganapathy

, Palanichamy

and Kannan

, An intelligent neuro-fuzzy temporal knowledge representation model for mining temporal patterns, Journal of Intelligent and Fuzzy Systems 26(3) (2014), 1167–1178.

12.

Sharma

, Shinghal

, Srivastava

and Singh

, Energy management for wireless sensor network nodes, International Journal of Advances in Engineering & Technology 1(1) (2011), 7–13.

13.

Anastasi

, Conti

, Francesco

D.M.

and Passarella

, Energy Conservation in WSNs: a Survey, Elsevier Science Publishers Ad Hoc Networks 7(3) (2009), 537–568.

14.

Sureshkumar

and Sabena

, Fuzzy-Based Secure Authentication and Clustering Algorithm for Improving the Energy Efficiency in Wireless Sensor Networks, Wireless Pers Commun (2020). https://doi.org/10.1007/s11277-020-07113-8.

15.

Maruthamuthu

and Sankaralingam

, Energy Aware Multiple Constraints Quality of Service Routing Protocol with Dynamic Mobility Prediction for Mobile Ad hoc Networks, Journal of Computer Science 7(6) (2011), 892–901.

16.

Al-Hunaity

M.F.

, Najim

Salam A.

and El-Emary

Ibrahiem M.

, A Comparative Study between Various Protocols of MANET Networks, American Journal of Applied Sciences 4(9) (2007), 663–665.

17.

Chen

and Nahrstedt

, Distributed Quality-of-Service Routing in Ad Hoc Networks, IEEE Journal on Selected Areas in Communications 17(8) (1999), 1488–1505.

18.

Zabin

, Misra

, Woungang

, Rashvand

H.F.

, Ma

N.–W.

and Ahsan Ali

, REEP: data-centric, energy efficient and reliable routing protocol for wireless sensor networks, IET Communications 2 (2008), 995–1008.

19.

Zhang

and Zhang

J.-h.

, A Self-adapted Anycast Routing Algorithm Based on Mobile Agent in WSN, Journal of Networks 6(2) (2011), 206–213.

20.

Selvakumar

, Sai Ramesh

and Kannan

, Fuzzy Based Node Trust Estimation in Wireless Sensor Networks, Asian Journal of Information Technology 15(5) (2016), 951–954.

21.

Zadeh

L.A.

, Fuzzy Sets, Journal of Information and Control 8 (1965), 338–353.

22.

Zadeh

Lotfi A.

, Fuzzy Logic=Computing with Words, IEEE Transactions on Fuzzy Systems 2 (1996), 103–111.

23.

Bede and Barnabas, Mathematics of Fuzzy Sets and Fuzzy Logic, Studies in Fuzziness and Soft Computing, Springer, 2013.

24.

Ren

, Liu

and Cao

, A parsimony fuzzy rule-based classifier using axiomatic fuzzy set theory and support vector machines, Journal of Information Sciences 181 (2011), 5180–5193.

25.

, Liub

and Li

C-B.

, An approach to expert recommendation based on fuzzy linguistic method and fuzzy text classification in knowledge management systems, Journal of Expert Systems with Applications 38 (2011), 8586–8596.

26.

Kamalanathan

, Lakshmanan

S.R.

, Arputharaj

, Fuzzy-clustering-based intelligent and secured energy-aware routing. In Handbook of Research on Fuzzy and Rough Set Theory in Organizational Decision Making, (2017), pp. 24–37. IGI Global.

27.

Selvakumar

, Karuppiah

, SaiRamesh

, Hafizul Islam

S.K.

, Mehedi Hassan

, Fortino

and Raymond Choo

K-K.

, Intelligent temporal classification and fuzzy rough set-based feature selection algorithm for intrusion detection system in WSNs, Information Sciences 497 (2019), 77–90.

28.

Thangaramya

, Logambigai

, SaiRamesh

, Kulothungan

, Kannan

, Ganapathy

, An energy efficient clustering approach using spectral graph theory in wireless sensor networks. In 2017 Second International Conference on Recent Trends and Challenges in Computational Models (ICRTCCM), (2017), pp. 126–129. IEEE.

29.

Sathish

, Ayyasamy

and Archana

, An intelligent beta reputation and dynamic trust model for secure communication in wireless networks. In Industry Interactive Innovations in Science, Engineering and Technology (2018), (pp. 395–402). Springer, Singapore.

30.

Asanambigai

, Ayyasamy

, February. Impartial Clustering Algorithm to Increase the Lifetime of Wireless Sensor Networks. In Intelligent Communication Technologies and Virtual Mobile Networks (2019), (pp. 453–459). Springer, Cham.

31.

Shi

, Wang

, Fang

, Gu

and Wang

, Wireless sensor network model with uncertain delay and packet loss based on intelligent fuzzy system, Journal of Intelligent & Fuzzy Systems 39(2) (2020), 1581–1590.

32.

Song

, Application of symmetry algorithm in energy optimal allocation of wireless sensor networks, Journal of Intelligent & Fuzzy Systems 38(6) (2019), 7715–7724.

33.

Van

N.T.

, Huynh

T.T.

and An

, An energy efficient protocol based on fuzzy logic to extend network lifetime and increase transmission efficiency in wireless sensor networks, Journal of Intelligent & Fuzzy Systems 35(6) (2018), 5845–5852.

34.

Sathiyavathi

, Reshma

, Parvin

S.S.

, SaiRamesh

and Ayyasamy

, February. Dynamic Trust Based Secure Multipath Routing for Mobile Ad-Hoc Networks. In Intelligent Communication Technologies and Virtual Mobile Networks (2019), (pp. 618–625). Springer, Cham.