1 Introduction

Wireless Sensor Networks (WSNs) are becoming more appealing for a variety of applications, including military reconnaissance, security surveillance, disaster management, habitat tracking, transportation and logistics, medical and health, factory automation, and so on, thanks to advancements and economic expansion in industry 4.0 technology and wireless communication technology [1, 2]. WSNs have therefore successfully established a link between the human society, computing world, and physical world. A WSN is made up of a huge number of small sensor nodes dispersed across a vast region, with one or more BSsor sinks to gather the data from these sensor nodes [3, 4].

Sensor nodes all have restricted power supplies, as well as information sensing, data processing, and wireless communication capabilities. Routing is one of the most significant technology in WSNs. Routing in WSNs is more complex than in typical ad hoc networks due to their fundamental features. To begin with, power, computational power, and transmission bandwidth are all rigorously inadequate. Second, creating a worldwide addressing system like the Internet Protocol is challenging (IP). Furthermore, IP cannot be utilized in WSNs due to the high expense of varying addresses in a dynamic or large-scale WSN. Finally, routing fails to deal with volatile and regular topology variations because of resource restrictions, especially in a mobile scenario [5].

Data aggregation by large number of sensors frequently leads to the data redundancy that increases the routing overheads. Because of many to one communication strategy in WSN, load balancing is required to improve the network lifetime. Finally, in time-constrained WSN applications, data transmissions should be completed within a certain length of time. As a result, restricted latency for data transmissions must be considered in these types of applications. Energy conservation is more important than quality of service (QoS) in most applications since all sensor nodes are restricted by energy, which is directly correlated to network longevity. WSN distributed over larger area needs clustering of sensor nodes in group to provide efficient routing and data aggregation. Selection of CH is crucial because it accepts the data from sensor nodes and transmits it towards BS. Many times the CH is selected based on its position in the cluster and often chances are given to centrally located node irrespective of its energy, load balancing capacity, connectivity, and distance from the base station [6, 7].

Various clustering approaches for WSN data aggregation and routing have been presented in the past. Janaki Raam et al. [8] discussed the impact of a parallel ACO algorithm and a k-means clustering approach for grouping sensor nodes in WSN routing to find the best path. It demonstrated a network with a long lifetime and a more efficient routing mechanism. An unsupervised clustering problem based on the ACO approach was proposed by Liu et al. [9]. They employed ACO and kept the best stochastic solution.—For effective CH selection, Gupta et al. [10] developed an improved version of the ACO base LEACH clustering method. In their study, data was sent from the node to the CH, then from the CH to the cluster leader, and finally from the cluster leader to the BS. As a result, energy consumption has fallen on average. They have not taken data redundancy into consideration, which has an impact on data aggregation. Clustering based ACO for VANETs (CACONET) was created by Aadil et al. [11] for clustering. They considered the network size, sensor nodes in the network, network coverage area, sensor node transmission range, VANETs node speed, and direction for algorithm testing. According to the results, the proposed technique outperformed the multi objective particle swarm algorithm (MOPSO) and clustering algorithm based ACO (CLACO). Yang et al. [12] suggested the ACO technique in combination with a dynamic clustering-based multipath routing protocol (MRP) for burst event monitoring in reactive WSNs in order to increase sensor network lifetime while reducing energy usage. CHs were chosen based on residual energy in the MRP algorithm, and numerous pathways between the sensor node and CH were chosen using ACO. It resulted in more efficient data aggregation, improved load balancing, increased sensor network lifetime, and reduced WSN energy usage. The authors discovered that selecting the parameters for the algorithm was difficult, and that parameter tuning resulted in slower speech.

Maheshwari et al. [13] presented cluster based routing based on butterfly optimization and ACO. Butterfly algorithm provides efficient clustering of the nodes and ACO provides optimal routing to transmit data towards CH and BS. It has given superior performance compared with LEECH and DEEC.Xiuwu et al. [14] investigated hybrid genetic Tabu search algorithm for the WSN routing to improve the network lifetime. It has given focus on the residual energy and node to BS distance for clustering that resulted in load balancing issue. Further, Yadav and Mahapatra [15] presented Cuckoo search based hierarchical routing to cope up with the energy stabilization and delay problem for cluster head selection to improve the networl lifetime. Selvi et al. [16] presented clustering routing in WSN based on cluster selection using gravitational approach and routing using clustered gravitational routing technique. It provided better network lifetime with reduced delay in data transmission. Rodríguez et al. [17] proposed Locust Search (LS) algorithm to determine the cluster heads and number of cluster heads for WSN clustering routing. Reddy et al. [18] have investigated the performance of ACO routing in collaboration with Glowwarm swam optimization (GSO) algorithm to deal with the energy depletion and network failure problem. Mehra et al. [19] presented zonal clustering in the heterogeneous network with mobile CH for the cluster head balancing. It provided better results than LEACH, and DEEC However, mobile nature of sensors needs higher energy consumption which limits the use of the system in extreme dynamic conditions. Various clustering and routing schemes has been presented to improve the lifetime, scalability, security, load balancing, and improve the disaster handling capacity in extreme environmental conditions [20,21,22,23]. Since, numerous optimization strategies are employed for the clustering routing in WSN, still the results are not optimized. There is need to focus on various performance parameters cumulatively such as network lifetime, delay, network overheads, load balancing, energy depletion, network stability, etc.

This paper present the energy efficient, load balanced WSN clustering and routing using ABC optimization algorithm and ACO (ACO) algorithm. The major contributions of this paper are summarized as follow:

  • Cluster head optimization using ABC (ABC) optimization based on variety of criteria such as cluster head load balancing factor, cluster head energy balancing, energy GINI coefficient, and inter and intra cluster distance.

  • Energy efficient routing using ACO (ACO) algorithm to improve the network performance, life-time, throughput and packet delivery ratio.

  • Novel Admission Allotment Scheme (AAS) to improve intra-cluster communication to minimize the overheads and packet drop.

The remaining paper is arranged as follows: In Sect. 2, the proposed energy-efficient load-balanced clustered routing protocol based on ABC and ACO is detailed. Section 3 contains the simulation details and comments of the suggested approach’s experimental outcomes. Section 4 concludes the essay with a concise conclusion and a future direction for the proposed system’s improvement.

2 Proposed methodology

The proposed method encompasses four phases such as network initialization, clustering, cluster head selection and optimization using FCM-ABC algorithm, and energy efficient routing using ACO algorithm. Figure 1 illustrates the flow diagram of the proposed system. The initialization phase includes network parameter initialization, radio model initialization, number of nodes, network area, initial energy, BS position, sensor nodes position. The radio model parameter initialization includes transmission energy for single bit, reception energy for single bit, amplification factor for multipath channel and free space, MAC protocol, traffic pattern, etc.

Fig. 1
figure 1

Flow diagram of proposed system

Full size image

2.1 Fuzzy C-mean clustering

The FCM algorithm splits the sensor nodes based on nodes position in the considered simulation area. It considers the homogeneous network where all nodes have same initial energy. The FCM provides the degree of membership of every cluster from the considered centralized cluster. The degree of sensor nodes nearer to centered node is higher and nodes away from centered node has lower value. The sensor nodes are grouped into cluster having higher membership function [24, 25].

Initialize the sensor node set as X = {x1, x2, x3 …, xn} and set of centers of clusters as V = {v1, v2, v3 …, vk}.

Step 1: Randomly select ‘k’ cluster centers.

Step 2: Calculate the fuzzy membership uij using Eq. 1.

$${\upmu }_{\mathrm{ij}}=1/\sum_{\mathrm{k}=1}^{k}({\mathrm{d}}_{\mathrm{ij}}/{\mathrm{d}}_{\mathrm{ik}}{)}^{(2/\mathrm{m}-1)}$$
(1)

Step 3: The fuzzy clusters ‘vj’ are calculated using \({\upmu }_{\mathrm{ij}}\) as given in Eq. 2.

$${\mathrm{V}}_{\mathrm{j}}=\left(\sum_{\mathrm{i}=1}^{\mathrm{n}}{({\upmu }_{\mathrm{ij}})}^{\mathrm{m}}{\mathrm{x}}_{\mathrm{i}}\right)/\left(\sum_{\mathrm{i}=1}^{\mathrm{n}}{({\upmu }_{\mathrm{ij}})}^{\mathrm{m}}\right),\forall \mathrm{j}=\mathrm{1,2},\dots ..\mathrm{k}$$
(2)

Step 4: Repeat step 2 and 3 until the minimum 'J' value is achieved or ||U(z + 1) – U(z)||< β.

Where, zdenotes for iteration step, U = (µij)n*c stands for the fuzzy membership matrix, β signifies the termination condition between [0, 1], and j defines the objective function.

2.2 Clustering optimization using ABC

The proposed method uses ABC algorithm for optimized CH selection from the initial FCMclusters derived using FCM. The proposed ANC used he is used for selection of optimized CHs from the clusters obtained using FCM algorithm. The improved ABC consider the cluster head load balancing factor, cluster head energy balancing, energy GINI coefficient, and inter and intra cluster distance for CH selection.

ABC is bio-inspired phenomenon invented by DervisKaraboga in 2005. It involves of three sets of bees: employed, onlookers and scouts [26, 27]. The employee bees discovers the food sources, onlooker bees decides the selection of food source and scout bees searches food in arbitrary direction when employee bees are abandoned. The count of employed bees is set same as the number of food sources around the hive. Employed bee whose food source finished becomes a scout bee. Figure 2 described the CH selection phenomenon of the ABC algorithm.

Fig. 2
figure 2

Flowchart of proposed ABC CH selection algorithm

Full size image

The ABC randomly generates potential SN solutions is equivalent to food sources.Let \(\mathrm{SN}=\left\{{\mathrm{S}}_{1}, {\mathrm{S}}_{2}, {\mathrm{S}}_{3},\dots \dots .{\mathrm{S}}_{\mathrm{C}}\right\}\) be the initial swarm population.

Equation 3 provides the probability function that Onlooker bees use for CH selection.

$$p_{i} = \frac{{F_{i} }}{{\sum\nolimits_{n = 1}^{SN} {F_{n} } }}$$
(3)

Here, \({\mathrm{P}}_{\mathrm{i}}\) is the fitness probability function generated by onlooker bees, Fi isfitness of ithsolution that is proportional to nectar quantity of the food source at positioni.

The Fi depends the cluster head load balancing factor, cluster head energy balancing, energy GINI coefficient, and inter and intra cluster distance for CH selection. The Eq. 4 provides the fitness function where weight factors must satisfy the condition given in 5.

$${F}_{i}={w}_{1}*{f}_{1}+ {w}_{2}*{f}_{2}+ {w}_{3}*{f}_{3}+ {w}_{4}*{f}_{4}$$
(4)
$${w}_{1}+{w}_{2}+{w}_{3}+{w}_{4}=1.$$
(5)

2.2.1 Fitness for energy GINI coefficient

The \({\mathrm{f}}_{1}\) provides the cluster head energy balancing based on the GINI coefficient that provides uneven distribution of energy in the cluster as given in Eq. 6. Basically, GINI provides the income distribution of the population [28].

$${E}_{s(G)}=\frac{1}{{2num}^{2}\left(s\right){E}_{ave}(s)}\sum_{i=1}^{num(s)}\sum_{j=1}^{num(s)}\left|E\left(i\right)-E(j)\right|$$
(6)

where \({\mathrm{E}}_{\mathrm{s}(\mathrm{G})}\) is energy Gini coefficient of the sth cluster, \({\mathrm{E}}_{\mathrm{ave}}\left(\mathrm{s}\right)\) isthe average remaining energy of the sth cluster, num(s) depicts the number of sensors in the sth cluster, \(\mathrm{E}\left(\mathrm{i}\right)\) signifies the residual energy of nodei.

The standard deviation of GINI coefficient is computed using Eq. 7. The lower value of \({\mathrm{E}}_{\mathrm{s}}\) says the energy balance is similar in cluster and can be used for form effective cluster.

$${E}_{\sigma }=\sqrt{\frac{\sum_{s=1}^{k}{(E\left(s\right)-{E}_{ave}0)}^{2}}{k}}$$
(7)

where \({\mathrm{E}}_{\upsigma }\) represents measure of the degree of distribution of \({\mathrm{E}}_{\mathrm{s}(\mathrm{G})}\) of k clusters, E(s) is the residual energy of the sth cluster, ksymbolized total CHs of the presentnetwork, \({\mathrm{E}}_{\mathrm{ave}}\) depicts the average residual energy of each cluster.

The fitness function corresponding to GINI coefficient is given in Eq. 8.

$${f}_{1}=\frac{{e}^{k}}{K}.{E}_{\sigma }$$
(8)

2.2.2 Fitness for cluster head energy balancing

The fitness function \({\mathrm{f}}_{2}\) corresponds to the cluster head energy balancing (\({\mathrm{f}}_{21})\) and cluster head energy proportion (\({\mathrm{f}}_{22})\) as given in Eqs. 9 and 10. The overall fitness function represents the energy balance degree and energy ratio of cluster heads as given in Eq. 11.

$${f}_{21}=1-{\left[\frac{1}{k}\sum_{s=1}^{k}(\frac{{E}_{CH}\left(s\right)}{{E}_{ave}\left(CH\right)}{)}^{1-\varepsilon }\right]}^{1/1-\varepsilon }$$
(9)

where ksymbolized total CHs in the network, \({\mathrm{E}}_{\mathrm{CH}}\left(\mathrm{s}\right)\) is the residual energy of the s-th cluster head, \({\mathrm{E}}_{\mathrm{ave}}\left(\mathrm{CH}\right)\) denotesthe average residual energy of the CH, and \(\upvarepsilon\) characterizes the disparity aversion parameter, which is 0.5. The lesser \({\mathrm{f}}_{21}\) depicts the betterenergy balanced between the CHs.

$${f}_{22}=\frac{\sum_{i=1}^{n}E(i)}{\sum_{s=1}^{k}{E}_{CH}(s)}$$
(10)

\(\mathrm{n}\) is the count of surviving nodes in the existing network, E(i) represent the total energy of the current network, and ECH(s) characterized the total residual energy of the CHs.

$${f}_{2}={\omega }_{1}{f}_{21}+{\omega }_{2}{f}_{22}$$
(11)

2.2.3 Fitness for inter and inter cluster distance

The fitness function \({\mathrm{f}}_{3}\) is related to sum of distance between all CHs \({(\mathrm{f}}_{31})\) and total distance between node to its corresponding cluster head \({(\mathrm{f}}_{32})\) as given in Eqs. 12, 13, and 14. The larger \({(\mathrm{f}}_{31})\) denotes that clusters are evenly distributed. The smaller value of \({(\mathrm{f}}_{32})\) indicates that distance between nodes and CH is lesser that is used to indicate the compactness of the cluster.

$${f}_{3}=\frac{{f}_{32}}{{f}_{31}}$$
(12)
$${f}_{31}=\sum_{s=1}^{k-1}\sum_{m=s+1}^{k}{D}_{CH}(s,m)$$
(13)
$${f}_{32}=\sum_{i=1}^{k}\sum_{j=1}^{num(i)}{D}_{CN}(i,j)$$
(14)

where \({\mathrm{D}}_{\mathrm{CH}}\) denotes the distance between sensor nodes and CH, \({\mathrm{D}}_{\mathrm{CN}}\) stands for inter-cluster distance.

2.2.4 Fitness for cluster head load balancing

The fitness function \({\mathrm{f}}_{4}\) is based on CH load balancing, which is calculated using Eqs. 15–18. The load balancing of cluster heads is crucial for WSN. Cluster heads must perform more operations and expend more energy than the rest of the group. As a result, reducing the load on the cluster head can boost network speed dramatically. The cluster head’s burden is proportional to the number of members in the cluster.

$${N}_{ave}=\frac{n-k}{k}$$
(15)
$${Th}_{max}={N}_{ave}+\frac{{N}_{max}-{N}_{min}}{k}$$
(16)
$${Th}_{min}={N}_{ave}-\frac{{n}_{max}-{N}_{min}}{k}$$
(17)
$${f}_{4}=(\frac{{N}_{max}-{N}_{ave}}{{N}_{max}})\frac{{N}_{h}-{N}_{u}}{k}$$
(18)

Here, \({\mathrm{N}}_{\mathrm{ave}}\) representsthe average count of sensor nodes in each cluster, \({\mathrm{N}}_{\mathrm{max}}\) and \({\mathrm{N}}_{\mathrm{min}}\) depicts the count of sensor nodes in the largest and smallest clusters, \({\mathrm{N}}_{\mathrm{h}}\) characterizes number of clusters having more count of sensor nodes members than \({\mathrm{Th}}_{\mathrm{max}}\), \({\mathrm{N}}_{\mathrm{u}}\) is the number of clusters having less count of sensor nodes than \({\mathrm{Th}}_{\mathrm{min}}\).

It produces new solution (position) from the old one in memory (position update equation for jth direction of ith candidate) using Eq. 19.

$$v_{ij} = x_{ij} + \varphi_{ij} \left( {x_{ij} - x_{kj} } \right)$$
(19)

where

  • ϕij (xij − xkj) is called step size and (i ≠ k)

  • k € (1,2,……SN)

  • j € (1,2,……D)

  • ϕij: random number within [− 1, 1]

If the position bees is remained un-updated for predefined duration (limit) then the food sourcexi is abandoned.

2.3 Energy efficient routing using ACO

Further, ACO algorithm is used to establish the energy efficient routing path from CHs to BS. The generalized flow chart of the ACO which consists of initialization of parameters such as the number of ants, initial pheromone value, pheromone evaporation rate, destination node etc. is shown in Fig. 3.

Fig. 3
figure 3

Generalized flowchart of ACO

Full size image

There are two types of ants in the artificial ACO algorithm, namely forward ants (FANT) and backward ants (BANT).

The forward ants start the movement from the nest to the destination for foraging the food. The forward ant carries the ID of the current sensor node, destination node and IDs of the neighboring sensor nodes and the link information that leads to them.It carries information about the routing table of the current node. The motive of forward ant is to find the destination node and memorize the route information. When the forward ant visits at the destination node, it is killed and it converts it to the backward ant. The process of forward ants is given in Fig. 4.

Fig. 4
figure 4

Flowchart of the forward ant system

Full size image

The backward ant starts the movement from the destination node towards the source node along with the memorized route of the forward ants. It also consists of the current node ID, destination node ID, IDs of neighboring nodes and connected link information. Backward ants update the routing table. While coming back to the source node backward ants lay down the artificial pheromone on the path. As soon as backward ant reaches to the source node, it updates the routing table and kills itself. The process of backward ants is given in Fig. 5.

Fig. 5
figure 5

Flowchart of the backward ant system

Full size image

2.3.1 ACO algorithm

Step 1: At steady intervals, from the source node, a forward ant k are sent to find a route until it reaches to the destination (all visited node is saved onto a memory Mk of ant).

Step 2: Transition probability for an ant k to move from i to j is given as by Eq. 20

$${p}_{ij}^{k}= \left\{\begin{array}{c}\frac{{\tau }_{ij}^{\alpha }{\eta }_{ij}^{\beta }}{\sum_{\mathrm{l\epsilon }{\mathrm{j}}_{\mathrm{i}}^{\mathrm{k}}}{\tau }_{ij}^{\alpha }{\eta }_{ij}^{\beta }}\dots \dots \dots ..j \epsilon {j}_{j}^{k}.\\ 0 \dots \dots \dots \dots else.\end{array}\right.$$
(20)

Here ant “k” at node “i”chooses the next node “j” by using Eq. (1). “j” is one of the neighboring node of “i”, ant “k” has more probability to select the next node with higher values of \(.\) where,

jik stands for list of non-visited nodes so that an ant k can skip visiting a node i more than once.

τij is the pheromone level of edge (i,j).

ɳij is the visibility of j when standing at i.

ά and β are pheromone level edge and visibility of node adjustment parameters (The value of ά and β lies between 0 to 1).

τ is the routing table which stores quantity of pheromone trail on the connection (i, j).

ŋ is the node visibility factor which can be given by \(\frac{1}{\mathrm{C }-{\mathrm{ e}}_{\mathrm{s}}}\) (Where, c is the initial energy level of the sensor nodes and es is the real energy level of sensor nodes).

Step 3: When a forward ant reaches to the destination node, it updates the pheromone trail level of the path to arrive at the destination and stores the path information in its memory and then transforms it into a backward ant.

Step 4: Before backward ant“k” starts its reverse journey, the destination node calculates and calculates the amount of pheromone trail that the forward ant k had dropped during its journey by equation 21.

$${\Delta \tau }_{k}= \frac{1}{( N- {Fd}_{k})}$$
(21)

where

\({\mathrm{\Delta \tau }}_{\mathrm{k}}\) is the amount of pheromone trail that the forward ant k will drop throughout its journey.

N is the total number of nodes.

Fdk is the distance traveled by the forward ant k which are stored in its memory.

Step 5: Whenever a node “i” receive a backward ant“k” coming from a neighboring node “j”, it updates its routing table by equation 22.

$${\tau }_{ij}^{k}=\left(1-\rho \right){\tau }_{ij}^{k}+ {\Delta \tau }_{k}$$
(22)

where ρ is an evaporation coefficient whose value lies between 0 ≤ ρ < 1.

Step 6: When the backward ant reaches the source node, the ant is eliminated and communication initiated with the best neighbors.

2.4 Admission allotment scheme for improved intra-cluster communication

The traditional clustering routing algorithms considers Time Division Multiple Access (TDMA) for the intra-cluster communication. Further, CH uses Carrier Sense Multiple Access (CSMA) scheme to broadcast the data. Once the neighboring node receives the request, it establishes the connection with the CH. The CH creates the TDMA slots equals to the number of members in cluster. Every node transmits the data in its allotted slot only. If the node do not have any data to send then its time slot goes unused that results in wastage of energy. The proposed Admission Allotment Scheme is based on merit based admission process in university/colleges where student having highest merit is allotted the university seat. In proposed AAS, once the CH (University) is selected, the members (students) send request to establish the connection with CH. The CH prepares the merit table of the nodes containing node ID (Enrollment Number/Roll Number), merit number, Eligible (Busy) or not Eligible (Free) status of the node. The merit table considers Eligibility status as 1 for the node with data to send and 0 to the node without data to send. The admission process is carried out in several rounds. In each round, the eligibility status of each node is updated. After creation of merit table, the CH gives chance to the node having 1 eligibility status and merit number based on residual energy. Nodes having high residual energy are given high merit number. Once the selected node goes out energy, it sleeps and give chance to the next sensor node in the queue for data transmission.

3 Simulation results and discussions

The proposed system is simulated using MATLABR2018b on windows environment using personal computer having 8 GB RAM and core i5 processor with 2.64 GHz speed. The network parameters and its specifications are given in Table 1.

Table 1 System and network parameters and specifications
Full size table

The results of the proposed FCM-ABC-ACO-AAS are compared with traditional clustering LEACH [29], LEACH-C [30] and FCM-DS-ACO [31] for the homogeneous network condition. Figure 6a–c illustrates the initial network scenario, FCM clustering and optimized cluster head selection using ABC algorithm. Figure 6d–h shows that the proposed FCM-ABC-ACO-AAS provides superior performance compared with LEACH, LEACH-C and FCM-DS-ACO.

Fig. 6
figure 6

a Network scenario for N = 100. b FCM clustering. c Clustering and CH optimization using FCM-ABC. d Packet transmitted to CH. e Residual energy. f Energy dissipation. g Packet transmitted to BS. h Packet transmitted to CH

Full size image

Table 2 provides the comparison of the proposed FCM-ABC-ACO-AAS algorithm based on round at which first node dead, last node dead to describe the network lifetime and energy efficiency. The experimental results shows the significant improvement in the lifetime over the traditional clustering approaches.

Table 2 Network lifetime performance comparison with previous approaches
Full size table

Recently, distinct machine and deep learning techniques are widely used for many computational science applications such as signal processing [32, 33], image processing [34, 35], data processing, etc. which have shown promising outcomes to improve the efficiency of the systems. However, because of unavailability of data and unpredictable surrounding conditions, use of these techniques is critical for WSN clustering and routing. Still, there is chance to utilize these algorithms for the automatic learning of WSN scenario to predict future condition for lifetime improvement og WSN [36, 37].

4 Conclusion

This research uses Fuzzy C-mean and an enhanced ABC Optimization method to provide optimal clustering and cluster head selection in WSNs to fix the challenges of low energy efficiency and short network lifetime. Longer network lifetime is aided by ACO-based energy-efficient routing. The ABC considers characteristics such as cluster head energy balancing, cluster head load balancing, energy GINI coefficient, and inter and intra-cluster distance for optimal cluster head selection. It has shown substantial improvement over the centralized CH selection utilizing the FCM approach in a variety of network density and scalability circumstances. The proposed admission allotment scheme provides the effective intra-cluster communication and minimizes the network overheads occurred during data transmission. The recommended algorithm is capable of giving greater performance in real-time applications due to its adaptability to changing surroundings.In future; the efficiency of the proposed approach can be investigated for the real-time scenario. The security aspect can be considered in future to improve the trust factor of WSN for reliable data transfer. Again, there is need to focus on minimization of computational complexity of algorithm.