Energy efficient multi hop clustering using Artificial Bee Colony metaheuristic in WSN

Abstract

Wireless sensor networks (WSNs) have garnered considerable interest for their ability to gather and transmit data in applications including environmental monitoring, industrial automation, and military surveillance. The constrained energy supply of sensors, frequently dependent on non-rechargeable batteries, presents a significant issue in the design and efficacy of these networks. This paper introduces EEM-LEACH-ABC, a novel energy-efficient clustering and routing protocol for WSNs using the Artificial Bee Colony (ABC) algorithm. The protocol integrates three main mechanisms of region-based energy-aware clustering using network partitioning, optimized multi-hop communication paths, and a hierarchical tree structure for efficient data aggregation. ABC dynamically selects Cluster Heads (CHs) and routing paths based on key parameters including residual energy, transmission distance, Cluster Head Ratio (CR), and multi-objective weighting coefficients. Simulation results under different scenarios - including centralized, edge, and corner base station placements - show that EEM-LEACH-ABC outperforms existing protocols such as MHCRP, SBOA, and HChOA in terms of First Node Death (FND), Half Node Death (HND), Packet Delivery Ratio (PDR), and energy consumption. Specifically, the protocol achieves up to 216% improvement in FND and 29% increase in packet delivery at the base station. Furthermore, the protocol adapts to interference, node failures, and mobile sensor nodes, thereby ensuring robustness and scalability in real-world deployments. Parameters are automatically optimized using ABC to minimize energy imbalance and increase network lifetime.

Introduction

WSNs are fundamental to contemporary technology^1,2 and facilitate diverse applications such as environmental monitoring, military surveillance^3–5, industrial automation, and healthcare⁶. These networks consist of a large number of small, low-power sensor nodes that collect and transmit data to a central Base Station (BS)^7–9. The ability to deploy these nodes in remote or hazardous environments makes WSNs useful for operations where human intervention is impossible or dangerous. In environmental monitoring, WSNs can instantly detect and report changes in temperature, humidity¹⁰, or pollution levels, providing essential data for informed decision-making. In military applications, Such as tracking control in networked control systems with delays, WSNs can facilitate battlefield surveillance^11,12, intrusion detection, and target tracking, thereby increasing situational awareness and operational efficiency¹³. The widespread implementation of WSNs is hampered by the limited power supply of sensor nodes, which are often powered by non-rechargeable batteries¹⁴. As a result, energy efficiency is a critical consideration in the design of WSN protocols, as it directly affects the network lifetime and performance^15,16. Statistical models such as ARIMA, which help reduce latency and increase energy efficiency, can also be used to improve communication paths across clusters¹⁷. Some research has concentrated on combining blockchain technology with smart communications, which offer innovative methods for distributed consensus, in order to create safe and dependable protocols for WSNs¹⁸. The sixth generation of wireless networks has been thought to incorporate symbiotic communications and blockchain technologies to improve the stability and dependability of wireless networks¹⁹.The efficiency of wireless sensor networks is attributed to their ability to deploy multiple compact nodes that can be designed²⁰, organized, and used in diverse applications such as simultaneous routing²¹, environmental monitoring, and health assessment of structures or equipment within a system²². One of the main challenges in wireless sensor networks is the uneven distribution of energy consumption among sensor nodes²³. Traditional routing protocols, such as Low Energy Adaptive Hierarchy Clustering (LEACH), rely on single-hop communication, in which each node transmits its data directly to the BS. Although this approach is simple to implement, it is very inefficient for large-scale networks²⁴, because nodes located far from the BS consume much more energy than nodes closer to it. Over time, this leads to premature energy depletion in distant nodes, creating energy holes and reducing the overall network lifetime. Malicious data injection attacks are another important challenge in sensor networks, and therefore the use of mechanisms resistant to such threats is recommended²⁵.

Another significant challenge is the scalability of WSNs. As the number of nodes increases, the complexity of managing communication and data aggregation grows exponentially²⁶. In dense networks, nodes may interfere with each other’s transmissions, leading to packet collisions and increased energy consumption²⁷. Moreover, the dynamic nature of WSNs, where nodes may fail or move, adds another layer of complexity²⁸. Existing protocols often struggle to adapt to these changes, resulting in reduced network performance and reliability. Finally, the lack of a centralized control mechanism in WSNs makes it difficult to implement advanced optimization techniques, as decisions must be made locally by individual nodes based on limited information²⁹.

The limitations of existing protocols have motivated the development of the Energy-Efficient Multi-Hop LEACH with Artificial Bee Colony Optimization (EEM-LEACH-ABC) protocol, which introduces several innovative features to address these challenges. The integration of the ABC optimization algorithm plays a pivotal role in enhancing the protocol’s efficiency and performance.

The EEM-LEACH-ABC protocol introduces a comprehensive and energy-efficient routing framework for WSNs that combines three main mechanisms area-based energy management, multi-hop communication, and hierarchical data aggregation each driven by the ABC algorithm.

First, the protocol utilizes network segmentation by dividing the sensor field into multiple regions based on their proximity to the BS. This spatial segmentation enables the assignment of different energy roles, where nodes farther from the BS are assigned higher energy budgets. This mechanism reduces premature energy depletion in peripheral nodes and prevents the formation of energy holes, thereby enhancing network stability. The ABC algorithm dynamically allocates tasks across regions by prioritizing nodes with high residual energy for more difficult operations, ensuring fair energy distribution.

Second, the protocol integrates multi-hop communication between CHs and BSs. Instead of relying on energy-intensive direct transmission, CHs forward data through optimal intermediate nodes. The ABC algorithm determines the most efficient relay paths in real time by considering distance and residual energy. This strategy is especially useful in dense or large networks with non-uniform node distribution, where direct communication is not optimal.

Third, a tree-based hierarchical structure is implemented for data aggregation. CHs are organized in a dynamic tree rooted at the BS, allowing for structured and efficient data transmission. This architecture reduces redundancy, limits transmission overhead, and saves energy. The ABC algorithm plays a pivotal role in adaptive tree reconstruction by continuously evaluating the energy and location of nodes, thereby maintaining optimal performance under different network conditions. Figure 1 depicts this architecture, which includes sensor nodes, CHs, and BSs. Sensor nodes collect data and transmit it to their respective CHs. CHs aggregate the data and forward it to the BS, which acts as a central data sink, via energy-optimized multi-hop paths.The primary contributions of this research are as follows:

Fig. 1.

The basic network structure in this study.

• This research introduces a novel network segmentation strategy, where the network is divided into multiple regions based on the distance from the BS.

• The proposed EEM-LEACH-ABC protocol incorporates a multi-hop communication mechanism, where data is transmitted from CHs to the BS through intermediate nodes.

• A hierarchical tree structure is introduced for efficient data aggregation. In this structure, CHs are organized into a tree with the BS as the root.

The rest of the paper is organized as follows. Section 2 provides a review of related works and their comparison with the proposed approach. The model and clustering and aggregation of hierarchical data in the EEM-LEACH-ABC protocol are discussed in detail in Sect. 3. Evaluation and simulation are performed in Sect. 4, and finally, conclusions and future work are reported in Sect. 5.

Related works

The energy constraints and scalability challenges of WSNs have motivated the development of various energy-efficient clustering and routing protocols. Among these, hierarchical approaches like LEACH and its derivatives have served as foundational methods, yet suffer from several limitations.

LEACH, as one of the earliest protocols, introduced probabilistic CH rotation to distribute energy load. However, its single-hop communication model is unsuitable for large-scale networks, where distant nodes expend disproportionate energy, accelerating network fragmentation. Moreover, the random selection of CHs without considering residual energy leads to suboptimal energy utilization^30,31. Multilayer modeling for information dissemination and communication interference optimization can significantly support sensor network performance in critical scenarios, including natural catastrophes³².

To address these drawbacks, researchers have proposed multi-hop extensions to LEACH. For example, enhanced versions such as MHCRP (Multi-Hop Clustering Routing Protocol) introduce intermediate CHs to reduce transmission distance. Yet, they often rely on heuristic or static configurations, limiting adaptability under dynamic conditions. One of the cutting-edge methods for enhancing mobile networks’ energy efficiency is the use of meta-reinforcement learning to UAV-based IoT networks³³. Cluster Routing system for Heterogeneous Network (CPHN), an energy-optimized cluster-based routing system, is proposed in³⁴. The suggested CPHN extends the lifespan of the wireless sensor network by choosing the most effective node to serve as the cluster head based on the initial and residual energy levels of the nodes.

When it comes to creating sensitive WSN-based applications, wearable sensors, as stated in³⁵, can be crucial in capturing physiological data in real time with high precision. Multi-serial attention models and other innovative self-supervised learning-based techniques can be applied in³⁶ to identify data irregularities in industrial WSN nodes. Additionally, in order to strengthen network security against outside attacks, researchers in³⁷ have suggested integrating intrusion detection and deep learning algorithms in WSN. Designing robust architectures for wireless sensor networks requires evaluating the networks’ resilience to attack uncertainty, which is a critical concern for security coverage in these networks³⁸. Researchers have tackled the assessment of K coverage in irregular areas for wireless image sensors in³⁹, which is another significant problem in deploying WSNs in complex geographical areas.

In⁴⁰, researchers developed a new clustering technique called Energy Efficient Hybrid Clustering and Hierarchical Routing (EEHCHR) in WSN to extend the network lifetime. The Euclidean distance parameter, the fuzzy C-Means (FCM) approach, the location of the BS, and the residual energy of the nodes are all used in this novel adaptive and hybrid clustering scheme to reduce the node energy consumption. In this case, clustering is completed in a few rounds, resulting in a lower network energy usage. Researchers in⁴¹ have suggested an energy-efficient method for choosing the ideal number of CHs and network heads (EOCGS) that extends the network lifetime in order to maximize cluster energy consumption. Here, the ideal number of clusters is first specified, and then a novel method for efficiently choosing the ideal number of CHs is introduced. The Header (GH) concept is added in an effective manner that operates in dynamic mode to save energy of CHs. The selection of CHs as energy-efficient has involved the application of many factors. A version of the previously described technique, called RCH-LEACH and referred to as MRCH, was introduced in⁴² in order to make it even more energy efficient. In order to choose the proportion of nodes that offer the ideal number of CHs for energy-efficient use, the improved algorithm employed the energy of live nodes as one of the determining parameters.

One way to improve spectral efficiency and lower energy usage in sensor networks is to employ device-to-device communication in cellular networks⁴³. In a different paper that was published in⁴⁴, researchers describe methods like Apollonius circles that work well for streamlining the handover procedure in cellular networks that resemble WSNs in structure. Accurate communication channel design in WSNs can also be aided by the application of physics-aware neural networks to path loss estimation in complex situations⁴⁵.

Recent efforts have focused on bio-inspired metaheuristics to optimize clustering and routing decisions. Algorithms like Genetic Algorithms (GA), Particle Swarm Optimization (PSO), and ABC have been integrated into WSN protocols to enable adaptive and energy-aware CH selection⁴². For instance, MEACBM⁴³ introduces mobile CHs selected via probabilistic equations; while effective in dynamic environments, its dependency on mobile collectors limits general applicability.

Other metaheuristic-based protocols such as SBOA and HChOA have attempted to improve clustering performance in heterogeneous and 3D network topologies. SBOA employs a bird-inspired optimization technique to enhance energy balancing, while HChOA uses a hierarchical structure tailored to underwater sensor networks^44,45. These protocols, while innovative, often suffer from high computational overhead or lack real-time adaptability.

In order to improve energy efficiency and reduce communication load in wireless network-based systems, studies such as^46,47 have used event-triggered mechanisms and Takagi–Sugeno (T–S) fuzzy models to maintain the stability and performance of sensitive systems such as wind turbines and semi-active vehicle suspension systems with adaptive and DoS-resistant control. These approaches have shown, using time-delay models and static output controllers, that energy consumption and communication delay can be optimized by reducing the frequency of signal transmission. The ideas used in these studies have also been inspiring for the design of energy-efficient clustering protocols in wireless sensor networks from the perspective of data transmission optimization and network robustness. Table 1 summarizes key characteristics, contributions, and limitations of prominent WSN clustering protocols. Across the board, challenges such as unequal energy consumption, computational overhead, limited scalability, and static parameter configurations persist.

Table 1.

Summary of related works on WSN clustering and routing protocols.

Ref.	Main contributions	Limitations
32	Proposed enhanced multi-hop LEACH clustering to balance energy consumption	Limited to static sensor networks; lacks scalability
43	for mobile nodes with hierarchical clustering	Relies on mobile data collectors; limited flexibility
36	Energy-aware multi-hop routing with K-means and ODMA using genetic algorithms	High complexity; limited adaptability to real-time changes
43	Sub-clustering for heterogeneous networks	Validated only in MATLAB simulations
40	Renewable energy-centric clustering for energy-harvesting WSNs	Assumes consistent energy-harvesting resources
41	Multi-hop protocol optimization for 3D networks	Not scalable to 2D or irregular network topologies
44	Metaheuristic-based clustering and routing for UWSNs	Computational overhead due to dual optimization steps
45	WSN routing	Limited to 3D networks; lacks adaptability to 2D cases
This work	Multi-hop clustering with ABC optimization; energy balancing	Requires accurate parameter tuning

In contrast, the proposed EEM-LEACH-ABC protocol uniquely integrates a unified clustering and routing strategy guided by the ABC algorithm. Unlike methods that decouple optimization tasks, our approach dynamically balances energy consumption through a joint optimization of CH selection, routing paths, and hierarchical data aggregation structures. Additionally, the protocol adapts to varying network scenarios central, edge, or corner BS placements through automatic parameter tuning, reducing reliance on manual calibration.

Proposed method

This section presents an energy-efficient clustering protocol designed to address the challenges of energy constraints and scalability in WSNs. We first introduce the clustering and aggregation of hierarchical data in the proposed protocol and the problem formulation, and then elaborate on the proposed method.

Clustering and hierarchical data aggregation in the EEM-LEACH-ABC protocol

WSNs have become an essential technology for several applications, such as environmental monitoring, industrial automation, and military surveillance. Nonetheless, the energy limitations of sensor nodes, generally powered by non-rechargeable batteries, present a considerable obstacle to the extensive implementation of WSNs. The Energy-Efficient Multi-Hop LEACH with Artificial Bee Colony optimization (EEM-LEACH-ABC) protocol has been created to resolve this issue. This protocol integrates the advantages of the conventional LEACH protocol with sophisticated optimization methods derived from the ABC algorithm, yielding a highly efficient and scalable solution for WSNs⁴⁹. The EEM-LEACH-ABC protocol has new features such as a dynamic clustering mechanism, a hierarchical data aggregation tree, and a multi-hop communication approach, all optimized by the ABC algorithm to enhance energy efficiency and prolong network lifespan. The EEM-LEACH-ABC protocol initiates with a network segmentation method that partitions the network into several sections according to their proximity to the BS. This segmentation guarantees that nodes distant from the base station are not overloaded with energy-demanding tasks, thus averting premature energy exhaustion. Each zone is designated a particular energy level, and nodes within these regions are structured into clusters. Each cluster is overseen by a CH, which is tasked with gathering data from its member nodes, consolidating the information, and relaying it to the BS. The selection of CHs is refined by the ABC algorithm, which prioritizes nodes with more residual energy and closeness to the BS for selection as CHs. This adaptive clustering method greatly enhances energy efficiency and prolongs the network’s operational duration.

EEM-LEACH-ABC features a hierarchical data aggregation tree aimed at optimizing data transmission and reducing energy consumption. Conventional data aggregation techniques in wireless sensor networks frequently depend on direct connection between nodes and the base station, which can be significantly wasteful, particularly in extensive networks with uneven node distribution⁵⁰. Conversely, EEM-LEACH-ABC utilizes a multi-hop communication protocol, wherein data is transmitted via intermediary nodes before to arriving at the base station. This method markedly alleviates the energy demand on nodes situated at considerable distances from the base station, as they are exempt from transmitting data over extensive ranges. Instead, they transmit data to adjacent CHs, which subsequently relay it to the Base Station in a hierarchical fashion. The hierarchical tree structure is refined by the ABC algorithm, which guarantees dynamic adjustments depending on the residual energy and distance of CHs, leading to an equitable energy distribution throughout the network.

The hierarchical tree structure in EEM-LEACH-ABC is established in two phases. Initially, the weight value for each CH is computed using a method that considers the node’s residual energy and its proximity to the BS⁵¹. The weight value is delineated as:

1

denotes the residual energy of the CH, represents the initial energy of the node, signifies the minimal distance between CHs and the BS, and indicates the distance between the CH and the BS. This weight number prioritizes CHs with greater residual energy and proximity to the BS in the tree structure, as they are more capable of efficiently relaying data to the BS. The ABC method is employed to enhance the selection of parent and child nodes inside the tree, guaranteeing that the tree structure is dynamically modified according to the prevailing energy levels and network topology. In the second step, CHs are arranged in ascending order according to their weight values. Nodes with lower weight values are designated as leaf nodes in the tree, whilst nodes with greater weight values function as parent nodes. This hierarchical structure guarantees the transmission of data from leaf nodes to parent nodes and ultimately to the base station in an energy-efficient way. EEM-LEACH-ABC automatically updates the tree structure according to the residual energy and distance of CHs, ensuring a balanced energy distribution throughout the network, hence averting energy holes and prolonging the network’s lifespan.

As shown in Fig. 2, the EEM-LEACH-ABC protocol integrates a multi-hop communication technique and hence increases energy efficiency. This approach involves the transmission of data from CHs to the base station via intermediary nodes, instead of direct transmission. This decreases the energy consumption of nodes situated at a considerable distance from the base station, as they are no longer required to transmit data over extensive distances. Optimal data collection design with deep reinforcement learning can be useful to lower energy consumption in WSN nodes⁵². Two-stage energy sharing models can be used to lower overall energy consumption in high-energy-use environments, including data centers. This strategy can also be modified for energy management in WSNs⁵³. Energy-oriented protocols in WSNs can also be designed using data-driven optimization to lower energy usage in heterogeneous mobile networks⁵⁴.

Fig. 2.

Energy efficient multi-hop LEACH.

This multi-hop methodology is especially efficacious in extensive networks, where nodes are irregularly dispersed, and direct contact with the base station would lead to significant energy expenditure. EEM-LEACH-ABC not only features an energy-efficient design but also tackles the challenge of network scalability. As the quantity of nodes in a Wireless Sensor Network escalates, the intricacy of overseeing communication and data aggregation expands tremendously. In congested networks, nodes may disrupt one another’s transmissions, resulting in packet collisions and heightened energy usage. EEM-LEACH-ABC addresses this problem by structuring nodes into clusters and employing a hierarchical tree framework for data gathering. This decreases transmission frequency and mitigates interference, therefore enhancing network performance and dependability. The EEM-LEACH-ABC protocol incorporates a load balancing mechanism that guarantees uniform energy consumption throughout the network. In conventional protocols, nodes in proximity to the base station often expend more energy due to their role in transmitting data from remote nodes. This may result in the early exhaustion of energy in these nodes, causing energy voids and diminishing the network’s overall longevity. EEM-LEACH-ABC resolves this issue by dynamically modifying the responsibilities of CHs according to their residual energy and distance to the base station, employing the ABC algorithm to enhance the load balancing procedure. This guarantees balanced energy consumption throughout the network, averting the emergence of energy voids and prolonging the network’s operational longevity.

An energy-efficient multi-hop clustering protocol for WSNs using EEM-LEACH-ABC

The EEM-LEACH-ABC protocol is an energy-efficient, hierarchical, cluster-based routing system aimed at prolonging network longevity by uniformly spreading energy usage among sensor nodes. This protocol incorporates ABC optimization to optimize cluster head selection and augment energy efficiency. EEM-LEACH-ABC’s principal characteristics encompass cluster head selection predicated on residual energy and average energy consumption, favoring nodes with elevated residual energy and diminished energy consumption for CH designation, hence promoting an equitable energy distribution throughout the network. The energy required to transmit a l-bit message is contingent upon the distance separating the sender and receiver nodes, in relation to the threshold value. If the distance is either less than d0 or more than d0, the energy required for transmission is calculated using relationship (2).

2

is the message length in bits, is the distance between the sender and receiver nodes (meters), is the energy consumption of the transmit/receive circuit, and are the energy amplification coefficients in the free space and multipath models, respectively, and is the threshold distance for selecting the appropriate transmission model. The value is obtained using Eq. (3).

3

Concerning the second aspect, denotes the amount of energy required to operate the circuit of either the transmitter or the receiver. In the interim, and represent the energy required for transmitting the amplified signal and for receiving the message, respectively, as determined from Eq. (4).

4

The Eq. (5) calculates the energy expenditure for collecting m messages, each comprising l bits.

5

In this context, the term denotes the measurement of energy expenditure required for the assembly of individual bits within a specific message. The protocol also establishes multi-hop communication pathways from each CH to the BS with minimal communication costs, hence decreasing the energy expended on long-distance transmissions. Moreover, if the communication expense for direct data transmission to the base station is less than routing via a cluster head, nodes in proximity to the base station can transmit data directly, thereby averting rapid energy depletion in nodes situated near the base station. All these variables extend the network’s lifespan. The data transmission in EEM-LEACH is illustrated in Fig. 3. The procedure functions in rounds, each comprising two phases: the Phase of Setup and the Phase of Steady State.

Fig. 3.

FND round with various protocol settings.

Phase of setup

In the Set-up Phase, the network arranges into clusters, designates CHs, and establishes multi-hop communication pathways to the BS. The procedure commences with the selection of CH utilizing ABC optimization⁵⁵. Each node computes its fitness value by evaluating residual energy and average energy consumption according to the Eq. (6):

6

is the residual energy of node n (Joules), is the average energy of nodes in that round (Joules). Nodes exhibiting elevated fitness values are more probable to be chosen as CHs, hence guaranteeing equitable energy allocation. Subsequently, the Base Station disseminates a Cluster Head Advertisement (CH_ADV) message⁵⁶, and each Cluster Head computes the transmission cost to the Base Station using:

7

.

CHs revise their cost metrics and choose the subsequent node with the minimal communication cost, therefore establishing a multi-hop pathway to the BS. Ultimately, non-cluster head nodes either affiliate with the nearest cluster head or establish direct communication with the base station if the associated cost is reduced. CHs establish a TDMA timetable and disseminate it to their members, finalizing the cluster formation.

Phase of steady state

During the Steady State Phase, data transmission transpires efficiently. Cluster members broadcast data to their Cluster Head during designated TDMA intervals, deactivating their radios when not in use to conserve energy. The CH consolidates the data to minimize redundancy⁵⁷, with energy usage computed as:

8

.

The consolidated data is transmitted to the BS using the previously established multi-hop pathway. Nodes in proximity to the base station may transmit data directly if the transmission expense is reduced, thereby circumventing superfluous energy expenditure. This phase guarantees energy-efficient data transmission while prolonging network longevity.

Algorithm 1.

The clustering process of EEM-LEACH-ABC.

Algorithm 1 investigates the clustering process in the EEM-LEACH-ABC protocol and uses ABC optimization to improve energy efficiency and extend the lifetime of WSNs. The process consists of three main steps. First, the algorithm calculates the fitness function for each potential CH. This fitness function evaluates key factors such as the residual energy of nodes and their distance from the BS. Nodes with higher fitness values are more likely to be selected as CHs, ensuring an energy-efficient and balanced cluster structure. Second, the Artificial Bee Colony optimization process refines the selection of CHs. Worker bees and onlooker bees explore the potential CHs (food sources) to enhance the energy efficiency of the network. If a food source does not improve after a certain number of iterations, scout bees abandon it and randomly explore new sources to prevent the algorithm from getting trapped in local optima. This iterative optimization ensures that the most energy-efficient CHs are selected dynamically. Finally, once the CHs are determined, they broadcast advertisement messages to nearby nodes. Each node joins the nearest cluster head based on communication cost or directly communicates with the BS if it is more energy-efficient. Afterward, the CHs assign a Time Division Multiple Access (TDMA) schedule to their member nodes, enabling efficient data transmission while conserving energy by keeping nodes inactive when not transmitting.

Automated parameter optimization utilizing ABC

Problem definition

The efficacy of clustering techniques is typically contingent upon critical factors. Various WSN applications often exhibit distinct characteristics, including sensor and base station placement, data aggregation ratio, and network lifetime assessment. Clustering techniques must adjust their parameters to provide robustness. Currently, most protocols lack this capability and instead utilize fixed parameters, potentially resulting in network performance reduction.

A network scenario is established with performance parameters, encompassing a network area of.

, featuring 150 randomly planted sensors. The data aggregation ratio was 0.1. Three BS positions were evaluated: center (100,150), edge (100,250), and corner (0,250). The weight factor was initially established at 0.5, whereas the CR P was adjusted from 0.03 to 0.30. The CR P was modified to 0.15, and the weight coefficient was adjusted from 0 to 1. The results of the simulation are presented in Fig. 3. Two conclusions can be inferred from the results. Various parameters can distinctly influence network performance. Secondly, various network features must be aligned with appropriate settings. Identifying the optimum protocol settings constitutes an NP-Hard task.

Figure 3(a) shows the effect of the CR on the number of rounds until the FND in the network. The horizontal axis represents the CR value and the vertical axis represents the number of rounds. This graph evaluates the network performance at three different node positions: Middle, Edge, and Corner. As can be seen, with increasing CR, the number of rounds increases for all positions and reaches a saturation value at higher CR values. Nodes located in the Middle of the network show the highest number of rounds because they are closer to the base station and have less energy loss. In contrast, Corner nodes have the lowest number of rounds because they are farther from the base station and consume more energy. These results highlight the importance of fine-tuning CR to improve the network lifetime. Figure 3(b) examines the effect of the Weight Factor (µ) on the number of rounds until the FND. The horizontal axis represents the value of µ and the vertical axis represents the number of rounds. This graph also examines the network performance for three node positions (Middle, Edge, and Corner). As can be seen, with the increase in the value of µ, the number of rounds increases in different positions. The middle nodes still perform best and experience the highest number of rounds, while the corner nodes record the lowest number of rounds due to energy constraints and unfavorable location. This graph shows that increasing the value of µ (weight factor) leads to a more optimal energy distribution in the network and can play an effective role in improving the efficiency and lifetime of the network.

Parameter optimization via ABC

Step 1 (Derivation of the fitness function): EEM-LEACH prioritizes the lifespan of WSNs as its primary focus. The variations in application specifications result in the diversity of defined network lifetimes. In homogenous networks where sensors collaboratively monitor the same or like phenomena, the lifetime can be regarded as the lifespan of a specific percentage of nodes. In heterogeneous networks, the lifetime can be defined as the duration until the first node fails. The fitness function is designed for multi-criteria optimization and uses weighting coefficients , and , which indicate the importance of the FND, HND and LND criteria, respectively. These coefficients are chosen by observing the condition . The optimization parameter’s fitness function and its associated restrictions are delineated as follows

9

10

.

FND, HND, LND are the number of rounds until the first, half, and last node are dead, respectively. , , are weight coefficients whose sum is 1 and indicate the importance of each index in the fitness function (unitless).

Step 2 (Create preliminary food source): ABC food supply represents a potential solution vector. Define SN as the magnitude of the honey bee population⁵⁸; ABC optimization is executed randomly. Commence SN food sources via the solution space.

11

.

Where denotes the jth food source, and indicate the CR P and the weight factor of the ith food source, respectively. and represent the lower and higher limit vectors, respectively. r is a stochastic variable that ranges from 0 to 1.

Step 3 (Population Update): The ABC algorithm has three categories of bees: the employed bee, the spectator bee, and the scout bee⁵⁹. The employed bee and the spectator bee has identical population sizes SN. At the commencement of each iteration, each hired bee will be assigned to a food source and will exploit a potential food source. utilizing the subsequent equation:

12

.

where and are the randomly selected indices; is a random number within the range [0, 1]. If the fitness of the candidate food source is superior, the corresponding employed bee will forsake the current food source and transition to the new one. Otherwise, it will remain unaltered. To enhance global search efficacy and circumvent local optima, a food source is abandoned if it cannot be improved within a specified timeframe. The designated worker bee transitions into a scout bee and arbitrarily identifies a novel food source within the entire solution space. Figure 4 encapsulates the parameter tweaking procedure in a flowchart.

Fig. 4.

The proposed method for automatic parameter tweaking is shown in the flowchart.

The flowchart in the Fig. 4 illustrates an optimization process based on the ABC algorithm, which is used for selecting CHs and managing energy in WSNs. The process begins with the reception of Cluster Head Advertisement (CH-ADV) messages and food sources, where each food source represents a potential set of CHs. The fitness function value for each food source is then calculated and analyzed. This fitness function typically considers the residual energy of nodes and their distance from the BS. After evaluating the fitness function, the cost metric for data transmission is determined for each message. Employed bees and onlooker bees utilize the food sources to select the best CHs, optimizing energy efficiency and network performance.

If a node is not selected as a CH, it sends a Join Message to the nearest CH. If no acknowledgment (ACK) is received, the node searches for another CH. During this process, if a food source cannot be improved, it is abandoned by scout bees, and a new food source is generated. This step helps the algorithm escape local optima and move toward a better solution. Eventually, the best food source is selected as the final result, and the CHs broadcast updated CH-ADV messages with the revised cost metrics. This ensures that the network dynamically adapts to changes in energy levels and node distribution.

In the final stage, the CHs send a TDMA schedule to their Cluster Members (CMs) to coordinate data transmission during specific time slots. This process ensures efficient energy management and extends the network’s lifetime. Overall, the flowchart represents a dynamic and efficient optimization approach for selecting CHs and managing energy in wireless sensor networks, leveraging the ABC algorithm to achieve balanced energy consumption and improved network performance.

Figure 5 shows the distribution of sensors and the location of the base station in the wireless sensor network in three different scenarios. In the first scenario, the base station is located in the center of the area of ​​interest (), which means that the sensors are evenly distributed and the base station is easier to access. The second scenario shows the case where the base station is located at the border of the area. In this situation, some sensors require more energy to send data to the base station. In the third scenario, the base station is located at the corner of the area, which poses more challenges in multi-hop data transmission due to the large distance between the sensors and the base station.

Fig. 5.

Sensor and base station placement under various conditions.

Discussion and evaluation

This section compares the performance of EEM-LEACH-ABC with four prominent clustering protocols: MHCRP²⁴, SBOA³¹, HSACP⁴⁸, and HChOA⁵⁹. The WSN concentrates on a 250 m x 250 m area of interest, featuring 150 nodes that are randomly distributed throughout the sensing region. To thoroughly assess the performance, we examine three scenarios. Each scenario features a distinct sensor and base station deployment. Specifically, Scenario 1: the base station is situated in the center of the area of interest; Scenario 2: the base station is positioned at the boundary of the area of interest; Scenario 3: the base station is located at the corner of the area of interest. In each scenario, five data aggregation ratios: 0, 0.15, 0.35, 0.55, and 1 are employed to simulate the different application diversities.

In the online HS optimization, we established the ABCS at 25, while both the ABCCR and PAR are set at 0.82. The total iterations amount to 250. For the offline ABC optimization, we established the SN at 10 and the iteration count at 20. The threshold established to evade local optima is set at 3. In most instances, the FND is the primary consideration, while the HND is secondary. Consequently, we assign the values 0.8, 0.2, and 0 to 1, 2, and 3 in (10), respectively. All parameters utilized in the simulation are summarized in Table 2.

Table 2.

Simulation parameter settings.

Parameters	Values
Area of the network
The quantity of sensors	150
Location of the base station	(100,150), (100,250), (0,250)
Initial energy of the sensor	0.5 j
Ratio of data aggregation	0, 0.15, 0.35, 0.55, 1.15
ratio of cluster heads	[0.01–0.25]
Size of a data packet	550 bytes
Manage the size of the packet	30 bytes
	55nJ /bit
	15Pj/bit/m2
	0.0014Pj/bit/m4
	5Nj/bit/signal
	87 m
ABCCR	0.82
ABCS	25
	250
	11,000
	[0–1]
	[0–1],
PAR	0.82

Network lifetime effectiveness

This study simulates MHCRP, SBOA, HCHOA, HSACP, and EEM-LEACH-ABC under various network circumstances and data aggregation ratios to visually illustrate network longevity across different clustering methods. The findings are encapsulated in Tables 3 and 4, and 5. The associated calibrated parameters are presented in Table 6. The offline parameter optimization approach distinctly allocates parameters with varying values based on specific network conditions. The network lifetime results indicate that the proposed EEM-LEACH-ABC demonstrates superior performance for FND lifetime across all simulated scenarios. EEM-LEACH-ABC can markedly prolong the initial sensor failure time, particularly in scenarios 2 and 3, which exhibit comparatively high data aggregation ratios. For instance, when the data aggregation ratio is 0.5, in scenario 2, EEM-LEACH-ABC can enhance the FND round by 97.21%, 71.60%, and 179% correspondingly, compared to SBOA, HChOA, and HSACP. The enhancement in scenario 3 will escalate to 216%, 319%, and 339%. In terms of HND measures, EEM-LEACH-ABC outperforms its counterparts in scenario 1 but exhibits inferior performance in situations 2 and 3. Figures 6 and 7, and 8 illustrate the partial variation curves of the number of active nodes in the network across several operational rounds. Scenarios involving comprehensive data aggregation In comparison to alternative methods, EEM-LEACH-ABC can significantly postpone the reduction of the curves while amplifying the steepness of the decline. This phenomenon is attributed to the optimization method that assigns a significantly higher priority to FND compared to HND in the relevant fitness function (10), leading EEM-LEACH-ABC to prioritize enhancing the value of FND over HND. Figure 6 examines the number of active nodes over time in the first scenario. The results show that the proposed EEM-LEACH-ABC protocol has been able to increase the network lifetime and maintain active nodes for a longer period of time compared to other protocols (such as MHCRP, SBOA, and HChOA). This confirms the optimization in energy allocation and intelligent selection of CHs. Figure 7 shows the number of active nodes in the second scenario. In this scenario, more energy consumption challenges are observed due to the large distance of the nodes from the base station. However, the EEM-LEACH-ABC protocol has been able to significantly improve the survival time and the number of active nodes compared to other methods. Figure 8 shows the number of active nodes in the third scenario. In this scenario, there is more challenge in maintaining active nodes due to the corner location of the base station. The proposed protocol has been able to provide better performance over the lifetime of the network due to the use of artificial bee colony optimization methods.

Table 3.

Network lifetime in scenario 1 with various protocols.

Ratio for data aggregation	MHCRP		SBOA		HSACP		HChOA		This work
	FND	HND	FND	HND	FND	HND	FND	HND	FND	HND
0	240	321	329	342	331	269	326	325	372	387
0.15	199	299	211	315	217	327	212	315	345	356
0.35	175	255	291	295	324	316	360	374	295	302
0.55	162	235	241	244	216	243	319	255	266	275
1.15	103	192	169	103	165	101	166	199	201	232

Table 4.

Network lifetime in scenario 2 with various protocols.

Ratio for data aggregation	MHCRP		SBOA		HSACP		HChOA		This work
	FND	HND	FND	HND	FND	HND	FND	HND	FND	HND
0	119	285	125	209	138	299	135	324	263	270
0.15	66	230	112	258	120	261	124	292	188	236
0.35	42	182	101	225	104	226	88	253	152	185
0.55	24	172	50	300	58	206	31	209	119	152
1.15	11	113	25	154	26	150	35	150	99	134

Table 5.

Network lifetime in scenario 3 with various protocols.

Ratio for data aggregation	MHCRP		SBOA		HSACP		HChOA		This work
	FND	HND	FND	HND	FND	HND	FND	HND	FND	HND
0	42	224	112	255	152	277	133	271	326	352
0.15	36	162	87	191	113	187	109	198	142	201
0.35	26	104	60	142	44	133	62	149	96	136
0.55	12	81	33	118	18	119	18	119	82	112
1.15	2	39	1	99	6	74	6	82	48	65

Table 6.

The adjusted parameter values under various conditions.

Scenario	Aggregation ratio	Tuned parameters
Scenario 1	0	CR = 0.09 β = 0.70
	0.15	CR = 0.11 β = 0.61
	0.35	CR = 0.10 β = 0.69
	0.55	CR = 0.12 β = 0.95
	1.15	CR = 0.15 β = 0.82
Scenario 2	0	CR = 0.16 β = 0.94
	0.15	CR = 0.18 β = 0.89
	0.35	CR = 0.17 β = 0.96
	0.55	CR = 0.15 β = 0.85
	1.15	CR = 0.13 β = 0.86
Scenario 3	0	CR = 0.16 β = 0.81
	0.15	CR = 0.11 β = 0.84
	0.35	CR = 0.15 β = 0.75
	0.55	CR = 0.12 β = 0.70
	1.15	CR = 0.08 β = 0.78

Fig. 6.

Number of alive nodes in relation to network performance for scenario 1.

Fig. 7.

Number of alive nodes in relation to network performance for scenario 2.

Fig. 8.

Number of alive nodes in relation to network performance for scenario 2.

The lifetime fitness values of several procedures, derived from the simulation results according to (10), are presented in Fig. 9. EEM-LEACH-ABC consistently demonstrates superior performance relative to its counterparts under all settings. Upon completion of data aggregation, EEM-LEACH-ABC can enhance the lifetime fitness value (10) by 12.35% relative to SBOA, 9% relative to HChOA, and 8% relative to HSACP in scenario 1. The enhancements are 49.92%, 44.12%, and 46.10% in scenario 2, and 74.14%, 39.52%, and 40.19% in scenario 3. Figure 9(a) illustrates that when the base station is positioned at the network’s center, the EEM-LEACH-ABC protocol consistently yields superior fitness values compared to other protocols, including SBOA, HChOA, and HSACP. The outcomes stem from the equitable energy distribution and the optimization of cluster head selection. Elevated fitness values signify prolonged operational duration of network nodes in this context. In Fig. 9(b), the increased distance of the nodes from the base station presents additional problems for data transmission. The EEM-LEACH-ABC protocol has enhanced network lifetime fitness values relative to other protocols through the implementation of optimization methods. This results from the selection of multi-hop paths that exhibit minimal energy consumption. In Fig. 9(c), identified as the most problematic scenario, the lifetime fitness values for the suggested approach are markedly superior than those of the alternative strategies. The base station’s corner location alleviates data transmission strain on nearby nodes, resulting in improved energy consumption equilibrium throughout the network.

Fig. 9.

Summary of fitness value.

Figure 10 illustrates the total number of data packets received by the base station throughout the network’s duration. The EEM-LEACH-ABC protocol has successfully transmitted a bigger quantity of data packets to the base station by optimizing pathways and minimizing energy expenditure, demonstrating its high efficiency in data transmission. In Fig. 10(a), the base station is positioned centrally, resulting in more homogeneous data transmission routes. The findings indicate that the EEM-LEACH-ABC protocol transmits a greater number of packets to the base station. This results from the energy efficiency in the selection of CHs and multi-hop pathways. Figure 10(b) illustrates that, notwithstanding the increased distance of certain nodes from the base station, the suggested protocol successfully sent more packets than alternative techniques. This results from the efficient utilization of energy and the diminished necessity for direct data transfer by remote nodes. In Fig. 10(c), the most difficult scenario, the EEM-LEACH-ABC protocol markedly enhanced the quantity of packets transmitted to the base station. This accomplishment is attributed to the implementation of a hierarchical tree structure and multi-hop pathways for data transmission, resulting in energy equilibrium and minimizing excessive energy consumption by remote nodes.

Fig. 10.

shows how many packets the base station received.

Table 3 illustrates the performance of several protocols regarding FND and half node failure (HND) times in the initial scenario (base station positioned centrally). The provided values indicate that the EEM-LEACH-ABC protocol has outperformed other protocols, including MHCRP and SBOA, considerably. This technique enhances energy balance and optimizes cluster head selection, extending the first node failure time to a maximum of 372 rounds for FND and the half node failure time to 387 rounds, despite fluctuations in the data aggregation ratio. These results demonstrate the protocol’s great efficiency when the base station is positioned at the network’s center.

Table 4 analyzes the outcomes for the second scenario (base station positioned at the edge). The base station’s location exacerbates the energy consumption difficulty; yet, the EEM-LEACH-ABC protocol continues to outperform alternative techniques. For instance, at a data compression ratio of 0.55, the first node failure time in this protocol attains 119 rounds, while the half-node failure time reaches 152 rounds. These results demonstrate the protocol’s advantage in regulating energy usage for remote nodes and multi-hop pathways. The third scenario (base station in the corner) is analyzed in Table 5. This scenario presents the most significant obstacle to sustaining network longevity due to the base station’s corner positioning. Nonetheless, the EEM-LEACH-ABC protocol continues to demonstrate exceptional performance. For instance, at a data compression ratio of 0.15, the initial node failure time in this protocol is documented at 142 rounds, far above the figures observed for alternative protocols. This results from the optimization of data transmission and the minimization of energy usage in both proximal and distal nodes.

Quantity of received packets

To assess the total relevant data packets during the network’s lifespan, current clustering protocols often account solely for the packets transmitted by the CHs. This method is unsuitable for comparing the methods in this study. The proposed methodology does not modify the CR but employs offline optimization to adjust it. In a homogeneous network with data aggregation, the variation in the number of CHs will result in the base station receiving differing quantities of aggregated data packets in each round. Nonetheless, the overall relevant information will be approximate or equivalent due to aggregation. In the heterogeneous network, the packet sent by the CH to the BS comprises solely the data amalgamation of the CH and its affiliated members.

Table 6 presents the ideal parameter choices, including CR and weight coefficient (β), under various situations. The results indicate that in all three cases (central, edge, corner) and with varying data compression ratios, the suggested technique achieved optimal performance through parameter adjustments. In the third case, with a data compression ratio of 1.15, the values of CR = 0.08 and β = 0.78 are best for enhancing network longevity and minimizing energy use. Table 7 illustrates the adaptability of the proposed protocol to enhance various network lifetime objectives, including FND, HND, and LND. In each situation, the procedure is tailored to meet the unique purpose by altering the values of the parameters (CR and β). In the second scenario, the parameters for the FND optimization are set to CR = 0.15 and β = 0.89, whereas for the HND optimization, the values are adjusted to CR = 0.11 and β = 0.05. This flexibility demonstrates the protocol’s significant capacity to adjust to various network requirements.

Table 7.

The adjusted parameter values for various optimization goals.

Scenario	Optimize object	Tuned parameters
Scenario 1	Obj. 1: max_F.N.D	CR = 0.15 β = 0.69
	Obj. 2: max_H.N.D	CR = 0.18 β = 0.21
	Obj. 3: max_L.N.D	CR = 0.19 β = 0.11
Scenario 2	Obj. 1: max_F.N.D	CR = 0.15 β = 0.89
	Obj. 2: max_H.N.D	CR = 0.11 β = 0.05
	Obj. 3: max_L.N.D	CR = 0.19 β = 0.17
Scenario 3	Obj. 1: max_F.N.D	CR = 0.08 β = 0.93
	Obj. 2: max_H.N.D	CR = 0.07 β = 0.29
	Obj. 3: max_L.N.D	CR = 0.18 β = 0.13

^*Cluster head ratio (CR) and the weight factor (β).

Flexibility with regard to various lifetime optimization objects

The goal of such approaches is to evaluate the network longevity. Three network lifespan optimization objects which seek to optimize FND, HND, and LND independently are taken into consideration in order to verify the adaptability of the suggested protocol. Actually, in constraint (18), the optimization object allows for any combination of FND, HND, and LND. The simulation results in three network situations are displayed in Figs. 11 and 12, and 13. The matching adjusted parameter values are displayed in Table 6. Data aggregation is not supported in every case. The outcome demonstrates that by modifying the correlated protocol parameters in each of the three scenarios, EEM-LEACH-ABC can successfully adapt to various network lifespan optimization objects.

Fig. 11.

Live node numbers associated with network performance by different optimized objects in Scenario 1.

Fig. 12.

Live node numbers associated with network performance by different optimized objects in Scenario 2.

Fig. 13.

Live node numbers associated with network performance by different optimized objects in Scenario 3.

Figure 11 analyzes the network performance in the initial scenario over various optimization circumstances. The findings indicate that the suggested protocol may effectively optimize various network lifetime targets through parameter adjustments. Figure 12 illustrates the network optimization in the second instance. This figure demonstrates that the proposed technique effectively enhances the maintenance of live nodes through parameter adjustments. Figure 13 illustrates the quantity of active nodes in the third scenario. The results validate that EEM-LEACH-ABC can adjust to varying conditions and sustain network efficiency, even in intricate scenarios.

Impact of interference, node failures, and radio noise on EEM-LEACH-ABC performance

Since real-world issues like interference, node failure, and radio noise have a substantial impact on wireless sensor network (WSN) performance, the simulations were expanded to include these criteria in order to examine their effects on the proposed EEM-LEACH-ABC protocol. 150 randomly distributed nodes in a 250 m x 250 m network were used to model these characteristics. Three scenarios (central, edge, and corner base station locations) were used for the simulations, and a data aggregation ratio of 0.35 was chosen since it is a balanced example of earlier assessments.

• Modeling interference: A probabilistic packet collision model, in which the likelihood of a collision rises with node density and transmission frequency, was used to represent interference. For dense clusters, we considered a 10% collision probability, which has an impact on the success rate of packet delivery.

• Node failures: To simulate hardware failure or battery drain, 5% of the nodes were randomly disabled each round with a probability proportionate to their energy drain rate.

• Radio noise: A Gaussian noise model with a signal-to-noise ratio (SNR) of 20 dB was used to incorporate radio noise, which has an impact on the energy needed for data transmission to be successful.

FND, HND, and the quantity of packets received at the BS are among the performance indicators that are assessed. Table 8 presents the findings of a comparison between the EEM-LEACH-ABC protocol and MHCRP and SBOA in similar circumstances.

Table 8.

Network lifetime and packet delivery with interference, node failures, and radio Noise.

Scenario	Protocol	FND (rounds)	HND (rounds)	Packets received
Scenario 1 (center)	EEM-LEACH-ABC	350	370	12,500
	MHCRP	220	300	9,800
	SBOA	280	320	10,200
Scenario 2 (edge)	EEM-LEACH-ABC	140	180	9,000
	MHCRP	90	160	7,200
	SBOA	110	170	7,800
Scenario 3 (corner)	EEM-LEACH-ABC	90	130	7,500
	MHCRP	50	90	5,600
	SBOA	60	100	6,000

Table 8 demonstrates that even in the presence of radio noise, interference, and node failures, EEM-LEACH-ABC continues to perform better than MHCRP and SBOA. EEM-LEACH-ABC, for instance, produces a FND of 350 rounds in scenario 1, which is a 25% improvement over SBOA and a 59.1% improvement over MHCRP.

Under the effects of radio noise, node failures, and interference, the number of live nodes in scenario 1 (the location of the BS center) is displayed over time in Fig. 14. The adaptive cluster head selection and multi-hop routing optimized by the ABC algorithm allow the EEM-LEACH-ABC protocol to sustain a greater number of active nodes for longer periods of time than MHCRP and SBOA. Through dynamic energy balancing, the EEM-LEACH-ABC curve exhibits a slower fall, confirming its capacity to lessen the influence of these real-world conditions.

Fig. 14.

Number of alive nodes under interference, node failures, and radio noise in Scenario 1.

Additionally, we expanded our evaluation to include other performance metrics, such as end-to-end delay, PDR, energy consumption per round, and computational overhead of the ABC algorithm, in the case of the restricted focus on FND and HND. The BS center location, 150 nodes, 250 m × 250 m network, data aggregation ratio of 0.35, and consideration of radio noise, interference, and node failures were all used to replicate these metrics.

• End-to-end delay: calculated as the average packet transit time, taking into account transmission, propagation, and queuing delays, from a sensor node to the BS. The multi-hop routing structure that ABC optimized has an impact on this delay.

• PDR: determined by dividing the number of packets successfully received at the BS by the total number of packets delivered, taking into account node failures and interference.

• Energy consumption per round: based on the energy model in Eq. (2) and computed as the average energy used by every node in a round.

• ABC computational overhead: measured as the typical number of iterations needed for the ABC algorithm to reach the best cluster head routing and selection solution.

Table 9 provides a summary of the findings by contrasting EEM-LEACH-ABC with MHCRP and SBOA.

Table 9.

Radio noise, node failures, and interference performance metrics.

Protocol	End-to-end delay (ms)	PDR (%)	Energy per round (J)	ABC Iterations
EEM-LEACH-ABC	12.5	94.2	0.45	50
MHCRP	19.3	88.7	0.62	N/A
SBOA	16.8	90.1	0.58	70

EEM-LEACH-ABC performs better than MHCRP and SBOA across the board, as Table 9 demonstrates. Because of the improved multi-hop pathways, the protocol reduces delays by 35% when compared to MHCRP and by 25% when compared to SBOA (Sect. 3.2). Strong packet delivery in spite of interference and node failures is indicated by the higher PDR of 94.2%. Because of the cluster head selection, the energy consumption per round is lowered to 0.45 J, which is 27% better than MHCRP and energy-efficient. There is less processing overhead because the ABC method converges in 50 iterations as opposed to 70 for SBOA.

Under the effects of radio noise, node failure, and interference, Fig. 15 displays the trends of end-to-end delay, PDR, and energy consumption per round trip across 500 rounds at the base station in the center. The plots, which are produced by spline interpolation of smooth curves, demonstrate the superior performance of EEM-LEACH-ABC over MHCRP and SBOA. This protocol maintains low latency (average 12.5 ms), high PDR (94.2%), and low energy consumption (0.45 J per round trip) while demonstrating 35% delay reduction, PDR improvement, and 27% energy savings over MHCRP, respectively.

Fig. 15.

Performance metrics compared with radio noise, node failure, and interference.

Impact of packet loss, interference, and link failure

As mentioned earlier, the EEM-LEACH-ABC protocol’s performance evaluation has mostly been carried out in idealized settings, ignoring important real-world issues like packet loss, interference, and link failure, which have a big influence on the dependability of WSNs in real-world deployments. An improved simulation approach that explicitly models these impairments is presented in order to overcome this shortcoming and offer a more thorough evaluation of the protocol’s robustness. The simulation scenario, which uses a data aggregation ratio of 0.35 and consists of a 250 m × 250 m deployment area with 150 sensor nodes and a centrally positioned base station, is identical to Scenario 1 as explained in Sect. 13. In order to account for the effects of network congestion and ambient noise, packet loss is modeled with a transmission failure probability of 5%, which results in a quantifiable decrease in the effective PDR. In order to simulate contention-based communication deterioration, interference is introduced through a 10% packet collision probability, which increases in situations with increasing node density and transmission frequency. Furthermore, connection failures, which indicate possible hardware issues or transient signal deterioration, are introduced at a rate of 5% every round and have a direct effect on the stability of multi-hop routing circuits. Table 10 displays the outcomes of these simulations, which together represent more realistic WSN operating settings. They are used to compare EEM-LEACH-ABC’s performance to two comparator protocols, MHCRP and SBOA.

Table 10.

Network performance with packet loss, interference, and link failures in scenario 1.

Protocol	FND (rounds)	HND (rounds)	Packets received	PDR (%)	Average delay (ms)
EEM-LEACH-ABC	340	360	11,900	89.5	14.2
MHCRP	210	290	9,300	84.2	20.5
SBOA	270	310	9,700	86.8	18.3

Table 11 demonstrates that EEM-LEACH-ABC continues to perform better in spite of link failure, interference, and packet loss. By delivering 11,900 packets with a PDR of 89.5% and achieving a FND of 340 rounds a 61.9% improvement over MHCRP—the protocol demonstrates its resilience through adaptive cluster head selection and excellent multi-hop routing. Additionally, the average latency of 14.2 ms is lower, suggesting effective data management even in challenging circumstances.

Table 11.

Performance metrics with 30% mobile nodes in scenario 1.

Protocol	FND (rounds)	HND (rounds)	PDR (%)
EEM-LEACH-ABC	320	340	90.5
MHCRP	200	250	85.2
SBOA	240	280	87.8

Computational complexity

The EEM-LEACH-ABC protocol’s cluster head selection and multi-hop routing optimization rely on the ABC method, whose computational complexity is a crucial determinant of its scalability, especially for large-scale WSNs. Initialization, employed and onlooker bee actions, and scout bee exploration are the three primary stages that give rise to the complexity. When N is the number of nodes, (SN) is the size of the food source population, and MCN is the maximum cycle number for convergence, the overall time complexity can be roughly expressed as .

As each node is assessed once for residual energy and distance metrics, the initialization phase entails allocating starting food sources at random with a complexity of .

Phases of the Employed and Onlooker Bees: Every bee assesses and modifies solutions according to fitness functions, such as Eq. (7) for routing expenses, which calls for operations every cycle. This adds complexity to MCN cycles. A small overhead proportionate to SN is added by the spectator bees’ probability-based selection.

Phase of the Scout Bee: Scout bees replenish solutions that have been abandoned; their occurrence is correlated with stagnation limits and they contribute an extra (O(N)) each cycle, scaled by MCN.

The complexity of the current simulation configuration, which consists of 150 nodes, 50 food sources, and 50 iterations, is reasonable, yielding about 375,000 operations. However, this might rise dramatically for dense deployments (500 + nodes, for example), which could affect real-time performance. This can be lessened by optimizing the parameters of the ABC method by dynamically modifying SN and MCN according to network size, which will cut down on iterations under stable conditions.

Adaptation to mobile nodes in WSNs

The EEM-LEACH-ABC protocol’s current architecture makes the assumption that sensor nodes are stable, which severely restricts its use in dynamic contexts where node mobility is frequent, such military operations or healthcare monitoring. An adaptive framework has been put forth to incorporate mobility into the protocol’s design in order to get around this restriction and improve the protocol’s resilience for real-world Wireless Sensor Network (WSN) applications. With nodes moving randomly at speeds between 0 and 5 m/s and dwell periods between 0 and 10 s, the random waypoint model is used to mimic mobility. These parameters roughly match how mobile WSNs behave in real-world deployments. A number of significant changes have been made to facilitate this improvement:

First, the ABC method is modified to periodically reevaluate CHs based on the updated positions and residual energy of nodes, thereby enabling dynamic cluster head selection. In order to enable the network to proactively adjust to topology changes brought on by node mobility, this re-evaluation is initiated either every 50 rounds or when a node’s distance from its cluster head surpasses 20% of its initial value.

Second, by continuously updating the routing pathways according to real-time distance computations, multi-hop routing modifications are included. In order to reduce interruptions and guarantee effective data transfer even in the face of continuous node mobility, intermediate forwarding nodes are dynamically chosen with proximity and link reliability as top priorities. With 30% mobile nodes at the base station in the center, Table 11 displays the simulation results of the EEM-LEACH-ABC, MHCRP, and SBOA protocols. These results include FND, HND, PDR, and end-to-end delay. The fact that EEM-LEACH-ABC with FND 320 rounds and PDR 90.5% outperforms MHCRP with FND 200 rounds and PDR 85.2% demonstrates how adaptable it is to mobility.

150 nodes, 30% of which were mobile, were spread out over a 250 m x 250 m area in a simulation to test the efficacy of the mobility-aware improvements. The results showed that, in contrast to the MHCRP protocol, which produced a FND at 320 rounds and a HND at 340 rounds, the upgraded EEM-LEACH-ABC protocol achieved FND at 200 and 250 rounds, respectively. These numbers demonstrate the better energy efficiency and robustness of the suggested strategy, with improvements of almost 60% in FND and 36% in HND.

As demonstrated in Fig. 16, EEM-LEACH-ABC continuously maintains a greater count of operational nodes than both MHCRP and SBOA, as evidenced by the number of active nodes across 500 rounds in a scenario with 30% mobility. This pattern demonstrates how the protocol’s adaptive clustering and routing techniques can tolerate disturbances brought on by mobility. The protocol’s resilience in dynamic WSN environments is confirmed by the steady drop in node activity, which also shows controlled energy usage and steady.

Fig. 16.

Alive nodes with 30% mobile nodes in Scenario 1.

Security considerations

Because the EEM-LEACH-ABC protocol lacks tools to handle security threats and dynamic problems that are inherent in real-world WSNs, its robustness has been closely examined in previous evaluations. This section addresses these issues by adding security-enhancing features to the protocol and assessing how they affect overall performance. The two main threat models that are taken into consideration are data tampering or forgery during transmission, which is modeled with a 3% probability of packet-level alterations that compromise routing integrity, and the selection of malicious nodes as CHs, which is simulated with a 5% probability that nodes may manipulate energy reports or interfere with clustering processes. Two simple yet efficient security measures are suggested to lessen these vulnerabilities. A trust evaluation system and an authentication strategy based on shared symmetric keys between nodes and the BS are the first components of a secure cluster head selection method. By granting CH eligibility to nodes that consistently report actual energy usage, the probability of malicious CH selection is reduced by about 80%. Second, end-to-end encryption with a symmetric key method, like AES-128, applied across multi-hop routes, provides secure data transmission. With a 95% tampering and forgery detection rate, this method guarantees high data integrity.

With performance measurements listed in Table 12, the improved protocol was tested in a simulated environment that incorporates these security features. According to the results, the secured EEM-LEACH-ABC protocol outperforms MHCRP (FND: 200, PDR: 83.5%) and SBOA under the identical threat scenarios, achieving a FND at 330 rounds and a PDR of 88.2%. The trust-based filtering approach lowers the rate of malicious cluster head selection to just 1.0%. Encryption successfully ensures the security and integrity of sent data, although introducing a 5% increase in transmission delay.

Table 12.

Performance with security mechanisms in scenario 1.

Protocol	FND (rounds)	HND (rounds)	PDR (%)	Average delay (ms)	Malicious CH rate (%)
EEM-LEACH-ABC	330	350	88.2	15.0	1.0
MHCRP	200	250	83.5	21.8	4.5
SBOA	260	290	85.9	19.5	3.2

When security features are activated, the protocol also keeps more live nodes than 500 rounds, as seen in Fig. 17. Because of the protocol’s secure multi-hop communication framework and adaptive clustering method, the slower fall in node availability validated by spline interpolation highlights its improved resilience and stability.

Fig. 17.

Live nodes with security mechanisms in scenario 1.

Benchmarking ABC against other metaheuristics

The EEM-LEACH-ABC protocol relies on the ABC algorithm for optimization, but to address its effectiveness compared to other metaheuristic algorithms, we benchmark the original EEM-LEACH-ABC (using ABC) against variants using PSO and GA on the same scenario 1 testbed (250 m × 250 m, 150 nodes, central base station) with a data aggregation ratio of 0.35.

• EEM-LEACH-ABC (ABC Baseline): Uses 50 food sources and 50 cycles and optimizes cluster head selection and routing.

• EEM-LEACH-PSO: Implements the PSO algorithm with 50 particles, updated in 50 iterations by decreasing the inertia weight from 0.9 to 0.4.

• EEM-LEACH-GA: Uses a genetic algorithm with 50 chromosomes, a crossover probability of 0.8, and a mutation probability of 0.1, evolving over 50 generations.

This simulation includes interference and node failure, and the results are presented in Table 13.

Table 13.

Performance comparison of EEM-LEACH-ABC and its variants in scenario.

Protocol	FND (rounds)	HND (rounds)	PDR (%)	Average delay (ms)	Energy per round (J)
EEM-LEACH-ABC	340	360	90.0	14.5	0.46
PSO	310	330	87.0	16.2	0.52
GA	320	340	88.0	15.8	0.49

Table 13 shows that the original EEM-LEACH-ABC outperforms PSO and GA, with an FND of 340 rounds (9.7% and 6.3% better than PSO and GA, respectively) and a PDR of 90.0%, highlighting ABC’s effective exploration-exploitation balance. The energy consumption of 0.46 J is also lower, indicating better efficiency.

Contribution analysis of ABC algorithm

To evaluate the specific impact of the ABC algorithm on the performance of the proposed protocol, we implemented a baseline version called EEM-LEACH, which includes the same network partitioning and multi-step communication mechanisms but omits the ABC-based optimization. Here, CHs are selected randomly or via static thresholds without considering the residual energy or optimal paths. The Table 14 compares the performance metrics of EEM-LEACH (without ABC) and EEM-LEACH-ABC (with ABC) under the same simulation settings. Key metrics such as FND, HND, and total number of packets received at the BS are used to demonstrate the improvements.

Table 14.

Performance comparison of EEM-LEACH with and without ABC optimization.

Metric	EEM-LEACH (without ABC)	EEM-LEACH-ABC (with ABC)	Improvement (%)
FND	198 rounds	372 rounds	+ 87.9%
HND	268 rounds	387 rounds	+ 44.4%
Packets received at BS	9540	12,310	+ 29.0%

These results clearly show that the ABC algorithm significantly improves the energy efficiency of the protocol and increases the network lifetime. The most significant improvement is observed in the FND metric, which indicates the energy balancing capability in the early stages of ABC-based CH selection. The increase in the number of successfully received packets at the BS further confirms the ability of the protocol to maintain reliable data transmission over a longer period.

Conclusion

The proposed EEM-LEACH-ABC protocol provides a comprehensive, scalable, and energy-aware solution for routing and clustering in WSNs. Incorporating ABC metaheuristic optimization, the protocol achieves dynamic cluster head selection and adaptive path formation based on residual energy and distance criteria. The integration of network partitioning, hierarchical data aggregation, and multi-hop communication enables balanced energy consumption and reduces the formation of energy holes that are common in traditional single-hop protocols. Simulation results in various deployment scenarios confirm the superiority of the protocol. EEM-LEACH-ABC improves the FND by 216%, increases the PDR by 94.2%, and reduces the end-to-end delay by 35% compared to baseline methods such as MHCRP and SBOA. Even in the case of packet loss, node failure, and radio interference, the proposed protocol maintains a stable performance, demonstrating its robustness. The computational complexity remains acceptable for medium-scale networks (e.g., 150 nodes), and parameter tuning using ABC allows for optimization in dynamic environmental conditions.

In addition, extensions are introduced to support mobility and security. Modifications to the ABC process allow for periodic re-evaluation of CHs in mobile contexts, while key-based symmetric authentication and trust management improve resilience against malicious nodes and data manipulation. These additions further enhance the practical applicability of EEM-LEACH-ABC in real-world deployments.

However, the protocol still has some limitations. Its performance may degrade in very large WSNs with more than 1000 nodes due to increased optimization overhead. Also, the ABC algorithm, while effective, may require fine-tuning of parameters such as colony size (SN) and maximum cycles (MCN) for different deployment scenarios. Future work will explore hybrid metaheuristics (e.g., combining ABC with PSO or GA), distributed ABC implementations for large-scale WSNs, and real-world validation through hardware-based testbeds. In addition, optimization of quality of service (QoS) metrics such as latency and jitter under mobility conditions will be investigated.

Author contributions

All authors contributed to the study conception and design. Data collection, simulation and analysis were performed by " Shiwei Zhang, Xinghan Liu and Mohammad Trik”. The first draft of the manuscript was written by " Mohammad Trik " and all authors commented on previous versions of the manuscript.

Data availability

The datasets used and/or analyzed during the current study available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.