Energy Oriented Multi-Objective Route Optimization Using Donkey Smuggler Optimization in Cluster Based Wireless Sensor Networks

WSNs are referred as combination of millions of sensor nodes which are dispersed in the environment for the purpose of sensing, assessing and receiving data. The nodes of WSN are cost efficient and have high capability to process and receive data [1]. The nodes present in WSN are positioned in a randomized distribution with a small autonomous device to detect the structure of the phenomenon in the physical conditions [2, 3]. WSN is widely employed in various applications, including manufacturing, national defense, traffic management, disaster avoidance, environmental monitoring, and health care applications. Even though the WSN has countless applications, it still has some limitations. These include limited transmission range, limited battery power, and limited storage. The non-rechargeable battery-powered sensor nodes will not last very long. Energy efficiency is therefore seen as WSN's primary goal. Implementing energy efficiency in the sensor nodes of WSN is a challenging task due to the presence of energy constrained sources which is operated independently for long time duration. Though, the non-rechargeable nodes can be activated by replacing the batteries which is a cost consuming process and has limited energy autonomy [4]. So, the base stations act as a significant component to send the data and process it.

Various energy efficient protocols are developed that offer an appropriate utilization of energy while transmitting the data packets. But the intruders present in the network leads to higher energy consumption when the data is transmitted for longer distance [5]. Considering the structure of the network, routing protocols are categorized into two classes such as flat routing protocols and cluster-based routing [6, 7]. The development of energy-efficient routing protocols aims to provide a suitable use of energy during packet broadcasting. Because they eliminate the energy waste that happens during collisions, overhearing, and idle listening, cluster-based routing protocols perform better than non-clustering protocols

The clustering-based routing is highly responsible to transmit data without energy loss and achieve the state of energy efficiency. In hierarchical design, the nodes with high energy capabilities are utilized to transmit information and the low energy nodes operates near the region of target [8]. The clusters are generated from splitting the sensors and one of the nodes present in the cluster is chosen as Cluster Heads (CHs). The CHs are responsible to gather data from the members of clusters and helps in transmitting data to the Base Station (BS) [9]. It is hard to select the CH from the applicable nodes of clusters due to its high energy consuming activity. So, replacement must be done at proper interval to maintain effective performance. But it is a time-consuming process and leads to lack of energy efficiency [10, 11]. Optimal clustering technique switches the one-hop communication among CH and sink node to alleviate the rate of energy consumption and enhance the network’s life time [12]. Additionally, routing plays a critical role to ensure a secured end-to-end transmission of data packets in WSNs. Energy awareness and minimized route maintenance is considered as the major requirement in the process of data transmission [13, 14]. So, this research focused in providing an energy-efficient cluster-based routing in the nodes of WSNs using optimization technique. 

The paper highlights the importance of energy optimization in WSNs, as sensor nodes operate on battery power and are often located in remote areas. The energy-efficient cluster-based routing technique using an EM-DSOA. This algorithm is inspired by real-world transportation behavior, including route searching and selection performed by donkeys. The technique aims to improve energy efficiency and enhance network lifespan in WSNs. The main contributions of the research:

1. Enhancing energy efficiency through K-Means clustering and optimal CH selection using EM-DSOA.

2. Making precise predictions of optimal paths to achieve energy efficiency with minimal data packet loss during transmission.

3. Performing energy-efficient routing using EM-DSOA by considering fitness parameters such as energy, communication cost, CH balancing factor, and node degree. The work in the context of existing research on energy-efficient protocols in WSNs, highlighting the need for more advanced optimal route selection with minimal data packet loss.

The proposed work utilizes EM-DSOA to develop the cluster-based routing to reduce the energy consumption and increase the lifespan of network. The position of the sensors, selecting the energy efficient CHs and generation of cluster-based routing is the important process while considering the implementation of EM-DSOA in selecting an optimal path. This method has three important phases named Clustering, CH Selection and Route Path Generation. The block diagram for the proposed EM-DSOA method is represented in Figure 1.

1.png

Figure 1. Block diagram for EM-DSOA method

The selection of Optimal CH and route is done using the proposed EM-DSOA to enhance the energy efficiency and improve the lifetime of WSN. The improvisation in the delivery rate is performed by optimizing EM-DSOA with the multi-objective fitness parameters such as energy, Communication cost, CH balancing factor and degree of the nodes.

3.1 K-MEANS clustering technique

The initialization of sensors is the foremost phase where the sensor nodes are deployed in a random manner over the large-scale networks like WSN. The nodes clustered using K-Means clustering technique, which divides the sensors into k number of clusters The K-Means clustering is mainly depend on calculating the Euclidian distance between the nodes. After clustering the network, then from each cluster optimal CH is selected using EM-DSOA

3.2 EM-DSOA method

In this phase, EM-DSOA is utilized to select the optimal nodes known as CHs. The optimal selection of CH helps to minimize the energy consumption rate and helps to enhance the lifespan of network. The major objective of EM-DSOA method is to achieve a proper fitness by evaluating energy, Communication cost, CH balancing factor and degree of the nodes. 

3.2.1 Multi objective fitness function for DSOA

EM-DSOA uses Multi Objective fitness function to select optimal CH with high residual energy. The fitness function considers parameters like residual energy, communication cost, balancing factor, and degree of nodes.

(1) Energy

Energy is one of the most significant fitness parameters to be considered in cluster-based method. According to the proposed technique, the allotment of CHs plays an important role in aiding overall efficiency. Whenever the data is transmitted in nodes with poor efficiency, it consumes more energy than the normal nodes. In the process of suggesting an effective CHs, the proposed technique considers the ratio of initial energy. Moreover, the proposed method makes the nodes energy dependent and helps the nodes to acts as an effective CH. The formula to detect the energy ratio in the fitness function is represented in Eq. (1).

$f_1=\sum_{i=1}^n \frac{E_o}{E_o-S(i) \cdot E}$            (1)

where $n$ is known as the total number of nodes and ${{E}_{o}}$ is the initial energy of the node. $S\left( i \right).E$ denoted the node’s energy at the present state. The ultimate aim is to lower the energy consumption of nodes to maintain a minimal energy consumption. Whenever the range of energy is reduced, the energy present in the node remains high. Selecting the higher residual energy nodes helps to enhance the performance and lifespan of the network. The direct impact is created whenever the node with minimal energy is selected as CHs.

(2) Communication cost

During the process of data transmission, the minimal power is utilized which is directly proportionated to the squared value of distance between the candidate nodes and the source nodes. The communication cost is evaluated using Eq. (2).

$f_2=\frac{D_{\text {Avg }}^2}{D_0^2}$            (2)

where the average distance present among the optimal nodes and the neighboring nodes is represented as ${{D}_{Avg}}$ and the radius of the transmission path is represented as ${{D}_{0}}.$

(3) CH balancing factor

The randomized sensor node organization in the network leads to formation of both bigger and smaller clusters which leads to imbalance in the network. So, CH balancing must be considered as a significant parameter to balance the energy consumption among the nodes. Eq. (3) denotes the formula to evaluate the balancing factor of CHs.

$f_3=\sum_{j=1}^i \frac{n}{i}-l_j$          (3)

where the total number of alive nodes are denoted as $n$ and the total number of cluster heads are represented as $i.$ The total number of sensor nodes present in the cluster is represented as ${{l}_{j}}.$

(4) Degree of the nodes

The node degree is the count of non-cluster head candidates who fits to the corresponding mobile nodes. Whenever there is minimal number of mobile nodes, the persistency occurs in a longer time duration. So, the node with minimum node degree should be considered and the node degree can be minimized using Eq. (4).

$f_4=\sum_{i=1}^m I_i$            (4)

Performing individual minimization leads to incompatibility so, it is better to combine the overall objectives utilizing weighted sum method. Additionally, every fitness parameter has different objective value so, normalization must be performed for every individual objective by assigning the weights as ${{\beta }_{1}},\text{ }\!\!~\!\!\text{ }{{\beta }_{2}},\text{ }\!\!~\!\!\text{  }\!\!~\!\!\text{ }{{\beta }_{3}},\text{ }\!\!~\!\!\text{  }\!\!~\!\!\text{ }{{\beta }_{4}}\text{ }\!\!~\!\!\text{ }$for each fitness function. Eq. (5) represented below helps to normalize the individual objectives.

$N=\frac{f_i-f_{\min }}{f_{\max }-f_{\min }}$            (5)

where the normalized value which lies in the range [0,1] is represented as$\text{ }\!\!~\!\!\text{ }N$ and the function value is denoted as ${{f}_{i}}.$ The minimal and the maximal value is represented as ${{f}_{min}}$ and ${{f}_{max}}$ respectively.

$F=\beta_1 \times f_1+\beta_2 \times f_2+\beta_3 \times f_3+\beta_4 \times f_4$            (6)

The formula to evaluate the overall fitness function is represented in Eq. (6).

3.2.2 Working of EM-DSOA based CH selection

EM-DSOA is initialized with a set of candidate nodes which are need to be chosen as CHs. The ultimate aim of the research is to detect an optimal path to transmit the data packets with minimum energy consumption. EM-DSOA is utilized in this research to detect the optimal path by considering the fitness function parameters such as energy, Communication cost, CH balancing factor and degree of the nodes.

EM-DSOA algorithm selects the CHs in two stages. A non – adaptive Smuggler algorithm finds the probable solution (i.e., location of CH). The smuggler algorithm verifies all the probable routes from the source to destination to effectively transmit the data without any intruders. In the part of smuggler algorithm, the fitness function is evaluated to detect the best path which helps the donkey to transmit the data effectively. The fitness of the smuggler algorithm is evaluated based on parameter such as energy, communication cost, balancing fact of cluster head and node degree. The smuggler algorithm determines the parameter for each solution by computing the fitness value using Eq. (1) as follows,

$f\left(X_i\right)=\frac{\sum_{j=0}^J X_{i j}+\prod_{j=0}^J X_{i j}}{\sum_{z=0}^Z X_{i z}+\prod_{z=0}^z X_{i z}}$            (7)

where the probable solutions are denoted as ${{X}_{ij}},$ the number of possible solutions is represented as $i.$ The number of parameters for each probable solution is denoted as $j$ and the inversely proportional possible solutions as denoted as $z.$ The parameters hold by the numerator and denominator is proportional directly and indirectly in a corresponding manner. The fitness solution (i.e., best cluster head) obtained from Eq. (7) will be organized in a group on the basis of their fitness value. Thus, the best function will be set up and transmitted to the donkey.

The donkey algorithm to is an adaptive routing algorithm that is used to provide better results in consistent traffics. To avoid Congestion and delay due to traffic, a choke packet is used to update the routing table, when the update is made in the routing table, the best solution will be updated in it. Whenever the choke packet lacks its fitness, the following steps are need to be followed:

Run: Select the path to find better solution and re-compute the best probable fitness value to upgrade the best solution. When the best solution is distinguished in non-adaptive phase is considered as worst, the drop is made in the non-adaptive phase to choose a best solution based on new changes.

Face and suicide: The best solution adopted in the first stage of the algorithm is no longer available due to the changes which affects the fitness. The solution of the first stage is dropped till it retains the ideal state and the solution in the second stage is utilized as a best solution. The best solution in the second stage is determined using the formula represented as follows:

best $_{\text {suicidesolution }}=f\left(X_i\right)-f\left(\right.$ best $\left._{\text {solution }}\right)$            (8)

where the number of probable solutions is denoted as $i.$

There is minimal fitness obtained in the solution populations so, the fitness of best solution is neglected from the solution of all probable solutions and the variation among them is considered as the newly obtained best solution.

Face and support: Whenever the overloading occurs in the best solution which was initiated by smuggler, the dropping is avoided in the solution by assigning the best solution of the second phase till the overloading is removed. In other words, two channels are utilized to transmit and receive the data to avoid re-evaluation of probable fitness solutions in the population. The next best solution used to avoid the overloading until the prior solution retains its normal form and it can be computed using Eq. (9) and the best support solution is evaluated using Eq. (10).

secondbest $_{\text {solution }}=f\left(\right.$ best $\left._{\text {solution }}\right)-f\left(X_i\right)$           (9)

best $_{\text {supportsolution }}$ $=$ best $_{\text {solution }}+$ secondbest ${ }_{\text {solution }}$           (10)

The second-best solution will be determined by Eq. (9) which is obtained by neglecting the fitness of the all-probable solutions. After this, Eq. (10) will be utilized to combine the best solution that uses two path ways to perform a single task.

The iterative process involved in EM-DSOA method which helps to find an optimal route to transfer the data with minimal energy consumption is described in the following steps,

-The process of reading takes place in the input data within the decision variable range of n1 and $\text{n}2$. Where, n denotes the dimensional solution of the matrix.

-The smuggler phase is initiated to create solutions in a random population for row =1 as n1 and column = 1 as n2.

The parameters of the random solution are created by using Eq. (11).

$\begin{gathered}\text { Parameters }(\text { row, col })=\text { rand }([\text { decision variable range }], n 1, n 2)\end{gathered}$          (11)

-Consideration is made as $e=1$ for $\text{n}1$ and the fitness value is evaluated for each solution based on Eq. (1), then the population is updated for all the probable solutions to set the best solution. The obtained best solution is passed to the phase of donkey algorithm.

-Compute the variation occurred while computing the fitness value. If any variation occurs, the fitness of the best solution will be probably low.

-Then according to the behavior of the donkey, the best solution is updated using the Eq. (8).

-After this, the solution with second best fitness value is considered as the best solution which is used to detect the optimal route for transmission of data packets.

-The solution with second best fitness supports the best solution in the population without updating it for population fitness which is represented in Eqs. (9) and (10).

3.3 EM-DSOA based routing

In the process of routing, the dimensions of every population is similar to cluster head quantity. This research aims to select an optimal route from every CHs to their corresponding Base stations. In DSOA, the route is created among the CH at the source and BS at the destination point. Moreover, the similar fitness functions used in cluster head selection is considered for optimal routing.

3.3.1 Initialization

In the process of routing, EM-DSOA presents the path to forward the data packet from Cluster head to the BS. The dimensional value of each solution of EM-DSOA is equal as the number of CH present in the network. Additionally, an extra position is included in the side of base station. Every best solution obtained from EM-DSOA algorithm is altered by a capable transmitted route among the node at source and destination. The evaluation of each probable solution is similar to the CHs present in their respective route of transmission.

3.3.2 Selection of optimal route

The proposed EM-DSOA considers the similar fitness parameters like energy, Communication cost, CH balancing factor and degree of the nodes which is utilized in cluster head selection to detect an optimal route for transmitting the data. The node at the source perform communication through Route Requests (RREQ) that gets transmitted to the nodes which is present in the neighboring position. Moreover, reverse routing is performed to transmit data to the source at the subsequent node. When the optimal path for routing is initialized, the CH at the source gathers data from the neighboring nodes. After this process, the transmission of data is initiated through the wireless sensor networks.

The overall flow and structure of the proposed EM-DSOA methodology

1. Initialization phase:

-Deploy sensor nodes randomly in the WSN.

-Use K-Means Clustering to divide the network into clusters.

2. CH selection using EM-DSOA:

-Smuggler algorithm (non-adaptive phase):

•Explore possible solutions (potential CH locations).

•Evaluate fitness based on energy, communication cost, CH balancing factor, and node degree.

-Donkey algorithm (adaptive phase)

•Use the best solution (CHs) for data transmission.

•Update routes dynamically based on fitness.

3. Route path generation:

-Optimal route from CH to BS is determined.

-Transmission of data packets from source nodes to BS through selected CHs.

4. Fitness parameters:

-Energy, communication cost, CH balancing, and node degree are used in both clustering and routing phases.

5. Data transmission and network optimization:

-Minimize energy consumption and improve network lifespan by selecting optimal routes.

The Donkey Algorithm receives the best solution found by the Smuggler Algorithm, but unlike the Smuggler, it can adapt to changing conditions during the transmission phase. If the selected path becomes inefficient (e.g., due to traffic or energy depletion), the Donkey Algorithm can adjust the route or switch to a backup path found by the Smuggler.

-Smuggler Algorithm explores all possible routes finds optimal path based on fitness.

-Donkey Algorithm uses the optimal path for adaptive data transmission adjusts dynamically to optimize performance in real-time. The Smuggler Algorithm focuses on the search and exploration of solutions, while the Donkey Algorithm handles the execution and adaptive optimization of data transmission over the selected routes.