RETRACTED ARTICLE: An Energy Efficient Framework for Densely Distributed WSNs IoT Devices Based on Tree Based Robust Cluster Head

Abstract

There is a developing effect of WSNs (wireless Sensor Networks) on genuine applications. Various plans have been proposed for gathering information on multipath routing, tree, clustering and cluster trees. Existing schemes can’t give an ensured dependable system to versatility, movement, and end-to-end association, separately. Such kind of problems to be moderate, the proposed scheme considers a densely distributed WSN system model related to Internet-of-Things (IoT) and tree based cluster formation depending upon sensor node deployment density. For each tree based cluster having one cluster head node to attain energy efficient data gathering, a reinforcement learning based fuzzy inference system (RL-FIS) will applied to determine the data gathering node for every cluster present in the densely distributed WSNs based on three metrics: neighbourhood overlap, bipartivity index and algebraic connectivity. We compare our proposed scheme with the other schemes. Simulation results indicate that our proposed scheme outperform the other schemes in overall energy consumption saving and prolong the lifetime of the network.

1 Introduction

The possibility of a widely interconnected, versatile, and dynamic ubiquitous processing condition has been proposed for quite a long time [1]. Only recently has WSNs innovation began to get acknowledgment as a key empowering strategy for the developing pervasive computing areas [2]. The combination of detecting and remote correspondence has prompted the improvement of WSNs on account of the adaptability, self-association, and the minimal effort of customary sensors. Because of their fast development, WSNs have been proposed for a plenty of utilizations, including ecological observing, fire detection, object tracking and body zone systems [3]. Such different application situations demonstrate that sensor systems screen generally extensive fields in addition dragging out the lifetime of WSN and ensuring parcel conveyance delays are basic for accomplishing adequate nature of administration [4]. They are particularly valuable for people when the checked field is hazardous and inaccessible. Hence, an extensive WSN is favoured and has been considered in the course of the most recent decade. Subsequently, commercial utilization of WSNs is relied upon to increment drastically in the exact not so distant future [5].

The difficulties of routing in WSNs are because of a few qualities that different it from alternate interchanges ad-hoc networks in wireless [6]. Because of restricted energy supply, a high pre-disposition and few sensor hubs in dense organization to disappointment are a portion of the difficulties. A few specialized difficulties should be considered in the design routing methodology [7]. The most imperative routing challenge is used in the energy consumption and organize lifetime because of the immediate effect of energy and in view of the energy compelled nature of WSNs [8]. Normally, the sending of sensor hubs is completed once and anticipated that would work for a long stretch. Consequently, during a routing procedure, it is essential for energy supply to the hubs to be adequate [9]. This will manage the hubs to work over a long term to do the transmission and routing of parcels. Notwithstanding sparing leftover vitality of hubs, adjusting energy utilization among sensor hubs likewise decrease energy utilization. Consequently, finding and choosing the ideal ways is basic [10]. Nowadays, applications of densely distributed WSNs are becoming a reality in the Smart Grid, the Machine-to-Machine (M2M) communications networks, smart environments, and Internet of Things [11].

In densely conveyed WSNs, topology control strategies will assume an imperative part in dealing with the information gathering unpredictability of such exceptionally muddled and appropriated frameworks through self-association abilities and they likewise need the capacities for flexibility and versatility to topology changes to guarantee legitimate coordination [12]. Accordingly, such frameworks would require productive conveyed topology control systems that will have the capacity to adapt to the inconsistency of the connections wasteful routes for example preservation of the restricted vitality supplies of battery controlled hubs. In perspective of topology control bunch based methodologies examined by analysts, the grouping approach develops a topology with various levelled structures that are versatile and easy to oversee [13]. The benefit of grouping is that a specific undertaking can be limited to an arrangement of hubs called bunch heads and they can be allotted for gathering, preparing and sending bundles from non-bunch heads. This system gives an effective system association [15]. Other attractive highlights of the grouping approaches incorporate the heap adjusting and information total or information pressure offered for delayed system lifetime [14].

WSNs based information gathering approaches utilizing mobile sing nodes (MSN) effectively fulfil this prerequisite to gather information from every one of the sensors in systems. However, existing inquires about on information gathering use cluster based, cluster head helped, Self-composed tree based and sink migration supported [15, 21] approach. Additionally, the non-clustering approaches deplete their energy rapidly on the grounds that the information activity on the hubs is heavier than on the others, which abbreviates the system lifetime. In spite of the fact that group head helped approach devour less energy than server—customer based models, the problem area issue can’t have stayed away from the point of view of long-haul utilize. Counting a mobile server (MS) in the system can take care of this issue and appropriate the heap adjust of hubs because that the MS changes its area for information accumulation. Notwithstanding, if a checked field is huge and numerous hubs are sent, it sets aside substantially more opportunity to gather information over the system than on an MSN-based model.

Due to the complexity of combining the routing strategy with compressive sensing, there are not more research efforts to put the routing strategy on data aggregation in achieving lifetime optimization. The main contributions of our scheme are as follows:

• 1. We propose a RL-FIS based scheme to determine the data gathering node for every cluster present in the densely distributed WSNs based on three metrics: neighbourhood overlap, bipartivity index and algebraic connectivity. Moreover, we elect robust cluster head based on residual power, distance and delay to communicate with mobile sink node to participate in the process of building a tree for every cluster.

• 2. Simulation results demonstrate that the proposed RL-FIS scheme outperforms existing schemes in terms of saving the overall energy consumption and keeping load balance.

The remaining paper is organized as follows. Section 2 discusses the literature work. In Sect. 3, we describe our proposed RL-FIS scheme in detail. Based on the proposed scheme, in Sect. 4, the parametric matrices analysis and experimental results of RL-FIS scheme is given. Finally, Sect. 5 gives the concluding remarks of this paper.

2 Related Works

Some of the recent research works related to the data gathering approaches for WSNs is listed below:

Dong et al. [16] have proposed an approach Reliability and Multi-way Encounter Routing for Large-scale WSNs. RMER was an information gathering approach. The commitments of the proposed approach are the accompanying: (a) non-hotspot regions were used to choose the more screen hubs and hotspot zones were used to choose the less screen hubs that were placed at near the Sink, which can prompt expanded system lifetime and occasion recognition unwavering quality. (b) The proposed approach sends information to the Sink by merging multi-way courses of event observing hubs into a one-way course to total information. Thus, energy consumption can be enormously diminished, in this manner empowering additionally expanded system lifetime.

Haseeb et al. [17] have proposed an adaptive energy for remote sensor networks, which was an aware cluster-based routing protocol. The fundamental focus of displayed adaptive energy aware cluster-based routing (AECR) protocol was used for enhancing information conveyance execution and energy protection. The proposed AECR convention contrasts from other energy effective steering plans in a few angles. Firstly, it creates adjust measured groups in view of hubs dissemination and stays away from arbitrary clusters arrangement. Also, it streamlines among inter-cluster and intra-cluster routing ways for enhancing information conveyance execution while adjusting information movement on developed sending courses and toward the end, with a specific end goal to diminish the unreasonable energy utilization and enhance stack circulation powerfully among hubs by abuse of system conditions, the part of Cluster Head (CH) was moved. In terms of different execution measurements, the AECR protocol beats state of the art was shown by the Simulation results.

Velmani and Kaarthick [18] have exhibited an effective cluster-tree based information gathering plan for extensive mobile WSNs. In WSNs the proposed Link-aware Cluster-Tree (VELCT) and Velocity Energy-proficient plot for information accumulation which would adequately alleviate the issues of versatility, scope separate, delay, tree, traffic, end-to-end association and force. In view of the cluster head area, the Data Collection Tree (DCT) developed by the proposed VELCT. The Data Collection Node (DCN) gathers the information bundle from the group head and conveys it to the sink because DCN in the DCT does not take an interest in detecting on this specific round. The composed VELCT plot limits the vitality misuse, lessens the conclusion to-end postpone and movement in group head by successful use of the DCT in WSNs. Simulation results have exhibited that VELCT gives good QoS as far as throughput, energy consumption, arrange lifetime and end-to end defer for portability based on WSNs.

Biason et al. [19] have described context-aware enhancement structure and an energy-centric, which was records for the effect of energy of the essential IoT functionalities framework and that returns along three fundamental specialized pushes: (1) adjusting signal-subordinate handling methods (pressure and highlight extraction) and correspondence tasks; (2) mutually planning channel get to and steering conventions to expand the system lifetime; (3) through the appropriation of reasonable learning models, giving self-versatility to various working conditions and of reconfigurable/adaptable protocols and algorithms. Further, they exhibited some preparatory outcomes that approved the viability of line of activity, and show channel get to methods permits an IoT system to progressively tune lifetime for flag bending and how the utilization of adaptive signal processing, as per the necessities managed by the application.

Nguyen et al. [20] have developed an energy harvested roadside IEEE 802.15.4 They considered the normally repudiating necessities for QoS provisioning in a coordinated way for an IEEE 802.15.4-based wireless sensor network (WSN). In particular, they considered variable activity states of IoT applications within the sight of an energy reaping system in view of the development of vehicles. To this end, they presented a versatile vitality effective calculation, alluded to as ABSD that adjusts the medium access control (MAC) parameters of IEEE 802.15.4 sensor hubs because of the queue occupancy level of sensor hubs and the offered movement stack levels. By adjusting the transmission parameters, the ABSD calculation limits the system dispute level which could thus enhance the EE and also the system throughput. Numerical results affirmed that proposed algorithms accomplish high EE and QoS values while broadening the lifetime of sensor hubs for open air applications.

Energy proficient information gathering is vital and trying for lifetime investigation in densely appropriated WSNs, because that it can prompt a sudden death of the system. The prior inquires about talked by techniques about the load balancing to alleviate energy productivity issue in densely disseminated WSNs, and proposed a different heuristic based answers for adjust the energy consumption in sensor hubs by changing their control of transmission.

The effective energy information gathering has additionally been considered in cluster-based thickly appropriated WSNs. The vast majority of the current works assume that energy gap situates around design energy-efficient routing protocols the sink to alleviate prolong the system and lifetime and the unequal energy consumption. Nonetheless, late examinations base on altered Expectation–Maximization (EM) strategy call attention to that energy gap does not generally develop near the sink and exceedingly relies upon some system parameters, for example, the energy consumption demonstrate, transmission scope of sensor hubs and number of information ask for messages noticeably affect the energy consumption of the sensor nodes. According to our recent literature investigation group head supported approach is appropriate to limit number of information request messages. Additionally, the portable sink based information gathering technique is adequate to broaden thickly disseminated WSNs lifetime.

Subsequently in this exploration we will intend to build up an efficient mobile sink construct information gathering strategies based with respect to robust cluster head selection algorithm for both sparing time and preserving system energy in densely conveyed WSNs particularly for IoT applications.

3 Robust cluster head aided data gathering in densely distributed WSNs for IoT Systems

Mobile Sink-Centric based information gathering approach is basic in densely disseminated wireless sensor systems (WSN), which for the most part causes higher energy consumption of the sensor hubs close to the sink hub, this issue is called energy-hole problem in WSN. In existing altered Expectation Maximization (EM) based clustering approach focused to choose an ideal number of clusters to limit the energy consumption. The issue in the current approach is absence of cluster head determination, when the span of the system and hub density increases, it naturally boost the energy consumption which prompts limit network life time. Consequently, beat the above said issue in this paper we are plan to propose a novel tree based cluster head supported vigorous information gathering plan which leads to maximize network lifetimes and in addition energy effectiveness. In proposed scheme each tree based cluster having one cluster head node for achieve energy effective information assembling, a RL-FIS will connected in the main stage to decide the robust cluster head for each cluster display in the densely dispersed WSNs in view of three measurements, for example, neighbourhood cover, bipartivity record and mathematical availability. At that point the dynamic system reconfiguration will be done by moving the area of sink hub and merge the cluster head hub, at whatever point hub disappointment obtain in any cluster. The whole framework will yield better execution as far as energy consumption and network lifetime. Then the adequacy of the proposed approach will be assessed by contrasting and the current methodologies.

3.1 System Model

Considering the above components we propose static and energy productive routing for a versatile and complex IoT. We considered the impact of utilizing our transmission algorithm over a thousand hubs conveyed in a 200 and 300 m² region with various quantities of hubs and found that static routing is extremely reasonable for versatile IoT applications. We have utilized the same layered system aside from the relay layer which isn’t utilized as a part of our system represents to the hierarchical network structure where all items put are static and follow the transmission based on static routing. Figure 1 demonstrates the mobile sink based wireless sensor model considered for IOT system application.

Fig. 1

Mobile sink based wireless sensor network

3.2 Robust Cluster Head Selection

Generally, the cluster head of the WSN is choose on the premise of parameters, for example, energy, delay and separation. As opposed to in IoT network, it is critical to consider the parameter of the IoT gadgets. Since, the WSN is related with the IoT gadgets; it is important to consider both the temperature and load of the gadgets. All in all, cluster head determination strategy based on the parameters, for example, vitality, delay, separation, temperature and heap of the IoT gadgets. Actually, the load, temperature, deferral and separation of the gadgets should be less and the energy should be higher.

• (1) Residual Power

In this paper, we consider the battery power level of every cluster node while calculating the residual power, in order to maximize the network lifetime. Whenever a node forwards a packet, it consumes certain amount of energy whose amount depends on factors such as the nature of packets, their size, access frequency, and the distance between the nodes.

The available power for node $x_{i}$, $RP(x_{i} )$ depends on the number of nodes for the cluster $i$. The maximum $RP(x_{j} )$ means the more steady power and the more energy power. Thus, the node with maximum $RP(x_{i} )$ is extremely likely to be selected as a cluster head and able to support the network lifetime for a long time. Therefore, we consider the residual power function to maximize the network lifetime, and it is defined by Eq. (1):

$$RP(x_{i} ) = \frac{{\sum\nolimits_{{x_{j} \in cluster\,_{i} }} {EP_{{x_{j} }} } }}{{n_{i} }}$$

(1)

where $n_{i}$ is the number of nodes in the cluster $i$, and $EP_{{x_{j} }}$ is the residual energy power of the node $x_{j}$.

• (2) Distance

The distance between the cluster nodes as well the mobile sink is computed using Eq. (2), where $dist(N_{v} ,N_{{v^{\prime}}} )$ computes the distance between the normal node and the base station such as mobile sink of the densely deployed sensor networks

$$dist(N_{v} ,N_{{v^{\prime}}} ) = \sqrt {(x_{{n_{v} }} - x_{{n_{{v^{\prime}}} }} ) + (y_{{n_{v} }} - y_{{n_{{v^{\prime}}} }} )}$$

(2)

Here $n_{v}$ is the every cluster nodes and $n_{{v^{\prime}}}$ is the mobile sink node and the formulation $dist(N_{v} ,N_{{v^{\prime}}} )$ is used to find out using Euclidean distance formula

• (3) Delay calculation

The delay value ranges from 0 to 1. The delay which is endured by the sensor nodes while the transmission of data to the mobile sink is determined using the equation,

$$_{{delay(N_{v} ,N_{{v^{\prime}}} )}} = \frac{{\mathop {S\left( {N_{v} } \right)}\limits_{q = 1}^{n} - S(N_{{v^{\prime}}} )}}{N}$$

(3)

The number of members under each cluster group should be less to remunerate the delay. $S(N_{v} )$ Denotes the signal strength of cluster nodes and $S(N_{{v^{\prime}}} )$ denotes the signal strength of mobile sink node in the particular network. $N$ represents the total number of sensor nodes.

3.3 Fuzzy Logic-Based Parameter Evaluation and Reinforcement Learning-Based Data Gathering Node Selection

The Cluster head hub chooses its information gathering hub. We consider three measurements particularly Neighbourhood overlap, Bipartivity file, and the Algebraic network.

Neighbourhood Overlap (NOVER) is the most direct metric to gauge the degree of shared neighbourhood between the end hubs of a connection. NOVER has been up to this point effectively utilized for group recognition and connections with a smaller NOVER score will probably connect two unique groups and connections with a bigger NOVER score will probably be between hubs in a similar group.

Bipartivity record alludes to each tree structure is said to be genuinely bipartite graph, if we could parcel the vertices of the chart into two disjoint sets to such an extent that every one of the edges in the chart are those that interface the vertices in one segment to vertices in the other segment and that here are no edges between vertices inside a similar segment.

The algebraic availability of a system is a quantitative measure of the robustness of the system as for interface expulsions. A network having a higher incentive for the algebraic network will probably remain associated after at least one link removals, and vice versa.

Fuzzy logic is utilized to join these three variables to lead an assessment on the moment reward of the choice (essentially the connection effectiveness from the cluster head node for the every neighbouring hub). In addition to this, we need to consider long haul reward of the choice too. All the more particularly, the decency of an information gathering hub determination is additionally subject to the activities of the neighbouring hubs. Here, we consider this by considering what amount does the information gathering hub close from the focal point of Cluster head Node.

3.3.1 Fuzzy Logic-Based Data Gathering Node Evaluation

• (1) Method:

The Cluster head node calculates the evaluation value for each neighbour as follows.

• A: Fuzzification

  Use predefined linguistic variables and membership functions to convert Neighbourhood Overlap (NOVER), Bipartivity Index (BI), and Algebraic Connectivity (AC) to fuzzy values.

• B: Mapping and combination of IF/THEN rules

  Map the fuzzy values to predefined IF/THEN rules and combine the rules to get the rank of the neighbour as a fuzzy value.

• C: Defuzzification

  Use a predefined output membership function and defuzzification method to convert the fuzzy output value to a numerical value.

• (2) Neighbourhood Overlap

We consider the Neighbourhood Overlap between the cluster head hub and the other neighbour hubs. Since an exact estimation of Neighbourhood Overlap is troublesome if certainly feasible in densely sent WSN condition, for straightforwardness, we select prompt neighbour hubs from cluster head gauge Neighbourhood Overlap. We characterize a Neighbourhood Overlap metric as

If $N_{CH} (u)$ represent the cluster head node and $N_{CN} (v)$ represent respectively the sets of neighbours of nodes

$$NOVER(u - v) = \frac{{2*\left| {N_{CH} (u) \cap N_{CN} (v)} \right|}}{{\left| {N_{CH} (u)} \right| + \left| {N_{CN} (v)} \right| - 2}}$$

(4)

The multiplication with 2 in the numerator is to include a typical neighbour hub the area of among the last hubs v and u. The denominator of this articulation the subtraction of 2 is to abstain from tallying hubs u and v as neighbours of each other and along these lines all the more precisely measure the degree of shared neighbourhood by just allowing for hubs in the areas of v and u other than v and u. In the event that both u and v have a similar arrangement of neighbours, will be 1 at that point NOVER (u − v). Then again, if u and v don’t have any basic neighbour, will be 0 at that point NOVER (u − v). Along these lines, the estimation will extend from 0 to 1 of NOVER (u − v): the bigger the esteem, the more prominent is the degree of shared neighbourhood.

• (3) Bipartivity Index (BI)

In a chart to measure the degree of bipartivity the idea of bipartivity index is used. From 0 to 1, the bipartivity file of a diagram ranges. In the event that a graph is genuinely bipartite, the Bipartivity index is 1 at that point and inside a similar segment there are no baffled edges between vertices. If a graph isn’t genuinely bipartite, at that point the bipartivity file will be under 1.

$$BPI(G) = \frac{{\sum\nolimits_{j = 1}^{n} {\cosh (\lambda_{j} )} }}{{\sum\nolimits_{j = 1}^{n} {\cosh (\lambda_{j} )} + \sum\nolimits_{j = 1}^{n} {\sinh (\lambda_{j} )} }}$$

(5)

If λ_j the eigenvalues of the adjacency matrix of a chart G of n vertices. To quantify the degree of shared neighbourhood of the end vertices of an edge in a diagram, we propose to process the bipartivity list on the egocentric system of the edge and utilize the supplement of the bipartivity record (1 − BPI) as the link stability score (LSS) for the edge.

• (4) Algebraic connectivity

The algebraic connectivity used a higher incentive by a network; it will probably remain associated after at least one connection evacuations, and vice versa. Laplacian framework of the systems second smallest eigenvalue is processed by the algebraic connectivity of a system.

$$L\left( {i,j} \right) = {\text{ }}\left\{ {\begin{array}{*{20}l} {D_{i} } \hfill & {for\;i = j} \hfill \\ { -\, A\left( {i,j} \right)} \hfill & {for\;i \ne j} \hfill \\ \end{array} } \right.$$

(6)

For a system of n vertices with adjacency matrix A (i, j) and degree vector D_i. The logarithmic availability of a system is a quantitative measure of the power of the system as for connect expulsions.

• (5) Fuzzification

Figure 2 demonstrates the fuzzy participation functions for the Neighbourhood Overlap, bipartivity index and algebraic network. The Neighbourhood Overlap participation function characterizes what degree has a place with {Bad, Medium, Good}. Essentially, the bipartivity list participation function characterizes what degree the algebraic connectivity belongs to {Light, Medium, Heavy} and what degree has a place with {Low, Medium, High}.

Fig. 2

Fuzzy membership functions (NOVER, BI, AC)

• (6) Mapping and combination of IF/THEN rules:

The cluster head for every cluster utilize the THEN/IF rules (see Table 1) to figure the rank of the taking an interest hubs just like an information gathering hub. Since there can be different guidelines applying in the meantime, the evaluation results are consolidated by using the Min–Max strategy.

Table 1 Fuzzy combination rules for data gathering node selection

• (7) Defuzzification:

For the defuzzification, we use the output membership function as shown in Fig. 3, and the Center of Gravity (COG) method where the xΨ coordinate of the centroid is the defuzzified value which shows the competency value of the node.

Fig. 3

Output membership function for the calculation of data gathering node competency value

3.4 Fuzzy based reinforcement learning

Reinforcement learning is one of the reputed algorithm, which is used in this paper to discover state and action pair value $Q\left( {i,a} \right)$ denotes the long-period expected reward for every state and action pair. The feasible state and action values represents the optimal policy that an agent intends to learn fuzzy combination rules.

Reinforcement learning algorithm steps

• 1. Set $Q(i,a) = 0,V(i,a) = 0;\,\forall i \in S,a \in A(i).$

• 2. Set $k = 0,k_{\hbox{max} } ,A = cons\tan t.$

• 3. Calculate initial state $i.$

• 4. Choose action $a$ based on fuzzy combination rules at state $i.$

• 5. Perform action $a$ and determine equivalent reward $r(i,a,j)$ and do following updates: $V(i,a) \leftarrow V(i,a) + 1,a = \frac{A}{V(i,a)}.$

• 6. Update Q-factor connect to state $i$ and action $a$ as:

  $$Q(i,a) \leftarrow (1 - \alpha )\,Q(i,a) + \alpha [r(i,a,j) + \lambda \max_{b \in A(j)} Q(j,b).$$

• 7. Set $k = k + 1,i = j.$ If $\le k_{\hbox{max} }$, back to the step iv, else go to the step 8.

• 8. Calculate feasible decision at every state $a^{*} (i)$ as:

  $$a^{*} (i) \in \arg \max_{b \in A(j)} Q(j,b).$$

In reinforcement learning algorithm steps, $Q(i,a)$ is state–action pair value, $V(i,a)$ is the number of selecting action $a$ at state $i$. $S$ is set of states and $A(i)$ is set of all admissible actions at state $i$. $k_{\hbox{max} }$ is learning iteration with maximum number that is identical to the number of fuzzy combination rules. Figure 4 demonstrates the flow diagram for Fuzzy based reinforcement learning algorithm.

Fig. 4

Flow diagram for fuzzy based reinforcement learning algorithm

4 Results and Discussion

In this area, under different parameter settings the proposed protocol is assessed by using the simulation results. The system test system was utilized to complete an execution investigation of proposed plan to contrast with LEACH and MBC. In these simulations, 100 homogeneous sensor nodes and 9 cluster-head nodes with unlimited battery energy [14] are placed in an area of 1000 × 1000 m². A stationary sink node (BSN) is placed outside the area of observation with unlimited energy. The power parameters used for the simulations are MicaZ, ZigBee application with 127 bytes packet size, IEEE 802.15.4 standard at the MAC and physical layer, linear battery model (1200 mAh) for sensor nodes, and two-ray signal propagation model. In a square locale, the size of information envelop is 512 bytes, within the cluster the transmission range 40 m, between the cluster 80–120 m the transmission run, 20 m is the detecting range, and the base station is situated in (x = 500, y = 1050). For each sensing node, the energy parameters can be set as 300 mJ.

Based on proposed data gathering scheme, the network performance was simulated in terms of the packet delivery ratio (PDR), throughput, delay, total energy, and speed. Figures 5, 6, 7, and 8 illustrate the relationship among the performance of the network (PDR, throughput, total energy consumption, and delay) and the number of deployed nodes. It is worth noting that LEACH, MBC fail to prolong the PDR, and HEED, total energy consumption, delay, and throughput as the number of node increases.

Fig. 5

Measurement packet delivery ratio with respect to number of nodes

Fig. 6

Measurement throughput with respect to number of nodes

Fig. 7

Measurement energy consumption with respect to number of nodes

Fig. 8

Measurement end to end delay with respect to number of nodes

In mobile sensor ambience, compared with LEACH, MBC, and HEEDthe proposed scheme has superior performance showed in Figs. 9 and 10. In the results of simulation, it can be articulated that proposed scheme has provided stable links and mended adapting to the high mobility environment. On high mobility environment, proposed scheme makes better PDR and less end to-end delay.

Fig. 9

Measurement bit error rate with admiration to number of nodes

Fig. 10

Measurement buffer occupancy with respect to number of nodes

However, proposed scheme improves for execution straight regardless of whether the quantity of sensor hubs increment over the system. In huge scale portability based WSNs, temperamental connections may cause the packet loss and retransmissions. In that case, it might build the energy consumption of sensor hubs. Moreover, it might diminish the PDR and throughput. The proposed plan can give stable connections and assurance the adjusted energy protection over the system. Accordingly, it can be reasoned that proposed plot is great at adjusting to the high portability condition.

Finally, it can be presumed that the proposed information gathering can help in sparing the sensor hubs leftover energy, expanding the system lifetime and enhancing system dependability. It is more ready to adjust to the high versatility condition with a superior correspondence quality.

5 Conclusion

The large-scale areas are monitor by using numerous sensor nodes with the growing impact of WSNs on real time military and civil applications. By using Cluster tree network management architecture is constructed, and is a proficient method. Exploit the network lifetime, throughput, PDR, stable link for mobile sensor nodes and residual energy are the ultimate goal. In this paper, robust cluster head aided data gathering scheme is proposed for densely deployed distributed WSNs. Each cluster selects a robust cluster head in terms of energy distance and delay. Then, the data gathering node is selected through cluster head based on neighbour nodes link efficiency in view of RL-FIS. The proposed scheme provides high stable link data gathering node selection and enhances the service metrics such as throughput, PDR, Bit error rate and End-to-End delay with a reduction of buffer occupancy related network traffic and provides minimal energy utilization than LEACH, HEED, and MBC by the experimental results.