Enhanced Aperiodic Threshold-Sensitive Stable Election Protocol (EATSEP) for WSNs

Abstract

Wireless sensor networks (WSNs) have emerged as vital technologies for safety-critical applications due to their flexibility, scalability, and reliability. However, existing models such as LEACH, SEP, and TSEP exhibit limitations in energy efficiency, stability, and adaptability to heterogeneous node conditions. To address these gaps, this research proposes a multilevel heterogeneity-based WSN model that optimizes cluster-head (CH) selection and energy utilization for enhanced network performance. Simulations were conducted in MATLAB under unequal energy level variations and compared with established protocols. Results demonstrate that the proposed model consistently outperforms existing approaches in terms of network lifetime, throughput, and energy efficiency. Statistical analysis reveals a best-case improvement of approximately 9000 rounds and a worst-case gain of about 3000 rounds when four heterogeneity levels are employed, compared to three levels. These findings highlight that both the degree of energy diversity and the distribution of energy nodes across levels are crucial for achieving optimal performance. Overall, the proposed architecture significantly enhances reliability, stability, and energy efficiency, making it well-suited for disaster management and other safety-critical applications.

1. Introduction

Natural disasters often occur in areas where deploying wireless sensor networks (WSNs) is challenging due to inaccessibility. These regions typically require power recovery solutions to ensure reliable data transmission. Such applications demand high reliability for timely reception of critical data, alongside an extended network lifetime, as recharging sensor nodes is both costly and impractical in remote areas. Furthermore, the increasing severity of natural disasters, exacerbated by global warming and environmental destruction, poses a growing threat to the world, particularly in underdeveloped countries where sophisticated disaster detection technologies are often unaffordable or unfeasible.

In light of these challenges, research efforts have been ongoing for over a decade to explore the potential of WSN technology in disaster management. However, many open questions remain, and further research is needed to address the full scope of challenges in this field. New advancements in WSNs could provide remarkable insights, which should be considered when designing networks for natural disaster management.

WSNs are inherently flexible and support multilevel nodes’ disparity by incorporating various types of sensors, such as temperature, displacement, chemical concentration, pressure, and noise sensors. Our proposed design protocol addresses the aforementioned challenges by incorporating a temperature-sensing approach where data is continuously sensed, but transmission occurs only when the temperature exceeds a critical threshold or after a predetermined number of rounds. This hybrid approach conserves energy, as data transmission consumes significantly more energy than sensing. There are two primary types of routing protocols for WSNs:

• Proactive Routing Protocol: In a proactive protocol, each node regularly maintains and updates its routing table. Transmission occurs continuously to ensure regular data reporting. This protocol is suitable for applications that require information on a regular, ongoing basis.
• Reactive Routing Protocol: In contrast, a reactive routing protocol only updates the routing table for data parameters that have changed significantly compared to previous values. Transmission occurs only when there is a substantial change in the sensed data. This makes reactive networks ideal for time-critical applications where updates are necessary only when drastic changes occur.

This paper proposes a hybrid routing protocol for wireless sensor networks designed to address the unique challenges of disaster management in remote areas. By optimizing energy consumption and ensuring reliable data transmission in time-critical scenarios, the proposed solution aims to enhance the efficacy of WSNs in responding to natural disasters. Although significant progress has been made in this field, continued research is necessary to further refine these protocols and ensure their practical application in disaster management systems worldwide.

The structure of the paper is as follows: Section 2 provides a general background of related research and ongoing efforts in the domain of WSNs for disaster management. Section 3 presents a detailed explanation of the designed protocol and its components. Section 4 discusses the simulation results and compares them with other existing schemes. Section 5 explores potential application areas for the proposed protocol in disaster management scenarios. Section 6 concludes the paper with a summary of the contributions and future research directions.

2. Related Work

Wireless sensor networks (WSNs) are composed of numerous sensor nodes, often numbering in the hundreds or thousands, which are typically deployed in an ad hoc manner to gather and transmit data to a centralized base station (BS). Depending on the specific requirements of the application, these nodes may communicate in various modes. WSNs are well-regarded for their versatility in supporting a wide range of monitoring tasks. In healthcare, a key application is the continuous observation of patients with hypertension, where timely medical intervention is crucial. However, the shortage of healthcare personnel makes continuous monitoring difficult, contributing to increased patient mortality. Recent progress in WSN technology, particularly the development of Wireless Body Area Networks (WBANs), has facilitated more efficient patient monitoring and alleviated some of the burden on medical staff. WBANs utilize small, energy-limited sensor nodes, making the design of energy-efficient routing protocols critical for extending network lifespan and ensuring reliable performance [1].

Wireless sensor networks (WSNs) are increasingly deployed in urban environments to address challenges in areas such as gas monitoring, traffic optimization, healthcare, disaster response, and security. In smart cities, WSNs enable real-time data collection and communication for applications like smart transportation, waste management, smart grids, and environmental monitoring. They are also crucial for disaster monitoring, supporting rapid earthquake detection and multimedia alerts. Recent approaches, such as a hybrid token ring MAC protocol with clustering and duty-cycle operation, have demonstrated low-delay, high-throughput transmission for diverse data types, highlighting the effectiveness of WSNs in urban and disaster-monitoring applications [2,3]. A number of energy-efficient clustering protocols have been proposed to extend the lifetime of wireless sensor networks (WSNs). These approaches focus on intelligent cluster-head (CH) election, particularly in heterogeneous environments where nodes differ in energy capacities. By distributing the CH role according to energy availability, they balance resource utilization, reduce redundant transmissions, and improve overall network quality of service (QoS).

One of the earliest and most widely used protocols is Low-Energy Adaptive Clustering Hierarchy (LEACH) [4,5]. LEACH assumes homogeneous nodes with equal initial energy, processing, and communication capabilities. CH roles are rotated to distribute energy consumption evenly, but the homogeneous assumption limits its practicality in real-world scenarios and large-scale deployments. Other well-known homogeneous wireless sensor network protocols include Power-Efficient GAthering in Sensor Information Systems (PEGASIS) [6] and Hybrid Energy-Efficient Distributed Clustering (HEED) [7].

To overcome this, several protocols were designed for heterogeneous networks [8,9,10]. The Stable Election Protocol (SEP) [11] extends LEACH by introducing two types of nodes (normal and advanced), where advanced nodes have higher energy. By assigning different CH election probabilities, SEP increases network stability and is particularly suitable for applications such as disaster management. The Enhanced SEP (ESAP) [12] further generalizes this approach to three levels of heterogeneity (normal, intermediate, and advanced), providing better reliability and energy balancing compared to LEACH and SEP. The Threshold-Sensitive SEP (TSEP) [13] also employs three energy levels but introduces a threshold-based reactive transmission scheme inspired by TEEN [14], where nodes transmit data only when sensed values exceed predefined thresholds. This saves energy in time-critical applications but is less suitable for scenarios requiring continuous data reporting.

Another influential protocol, the Distributed Energy-Efficient Clustering (DEEC) [15], selects CHs based on the ratio of each node’s residual energy to the average energy of the network, giving higher-energy nodes greater chances of becoming CHs. This approach significantly improves stability and lifetime compared to LEACH and SEP. Extensions such as Enhanced DEEC (EDEEC) [16] and Threshold DEEC (TDEEC) [17] refine this by incorporating three levels of heterogeneity and residual-energy-based thresholds, further balancing energy usage and extending lifetime.

Other works refine these strategies. Energy-Aware TSEP (EATSEP) [18] combines initial and residual energy for CH election while reducing long-distance transmissions through CH-to-CH routing. Improved TSEP (ITSEP) [19] enhances CH regulation by considering residual energy, node density, and distance to the base station, yielding significant gains in stability and efficiency. Hierarchical and zonal schemes have also emerged: Hierarchical SEP (HISEP) [20] employs dual-CH clustering with Voronoi partitions, enhancing load balancing in both 2D and 3D deployments; Improved Zonal SEP (IZ-SEP) [21] divides the network field into zones and balances CH election using residual energy and node density; and the Zone and Energy Threshold protocol (ZET) [22] uses zonal clustering with energy-aware thresholds, achieving superior performance over LEACH and SEP in both homogeneous and heterogeneous settings.

Overall, clustering algorithms have proven to be an effective approach for energy conservation in both homogeneous and heterogeneous wireless sensor networks (WSNs) [4,5,23,24,25]. These protocols demonstrate the progression from homogeneous clustering (LEACH) to heterogeneous, multilevel, and zonal clustering approaches (SEP, DEEC, EDEEC, TDEEC, and their variants). By incorporating residual energy awareness, threshold-based transmissions, and hierarchical designs, they collectively enhance energy efficiency, stability, and throughput, making them better suited for real-world WSN deployments.

3. Proposed Methodology

The Enhanced Aperiodic Threshold-Sensitive Stable Election Protocol (EATSEP) is developed to overcome the limitations of the Threshold-Sensitive Stable Election Protocol (TSEP) proposed by the study of [13]. EATSEP is designed as a hybrid routing protocol that integrates both proactive and reactive routing mechanisms. This dual capability enables efficient handling of periodic data transmission as well as time-critical event-driven communication, making the protocol particularly suitable for safety-critical applications. In our analysis, we adopt the radio model presented in [5]. As shown in Figure 1, this radio energy dissipation model defines the transmission energy for sending a 1-bit message across a distance d as follows:

$$E_{TX}(l,d)=\left\{\begin{array}{c}l{E}_{elec}+l{ϵ}_{f_s}{d}^{2}d<d_{\circ }\hfill \\ l{E}_{elec}+l{ϵ}_{mp}{d}^{4}d\ge d_{\circ }\hfill \end{array}\right.$$

(1)

where $E_{elec}$ is the energy required to run the transmitter or receiver circuit per bit, $d_{\circ }=\sqrt{{\displaystyle \frac{{ϵ}_{f_s}}{{ϵ}_{mp}}}}$ is the threshold distance at which the model switches from free-space to multipath fading, ${ϵ}_{f_s}$ and ${ϵ}_{mp}$ are the energy amplification factors for free-space and multi-path models, respectively. Equation (1) is derived based on a simplified radio energy dissipation model commonly used in WSN literature, represented in Figure 1. The model assumes that during data transmission, a sensor node consumes energy in two components: (i) electronics energy $\left(E_{elec}\right)$—the energy required to operate the radio electronics for functions such as modulation, coding, and signal processing; (ii) amplifier energy $\left({ϵ}_{mp}\right)$—the energy used by the transmitter’s power amplifier to overcome signal attenuation over distance. The following assumptions were made in the energy dissipation model:

• Nodes are stationary and have identical radio characteristics.
• The communication channel follows either free-space $\left(d^2\right)$ or multipath $\left(d^4\right)$ path-loss depending on distance.
• Power control adjusts transmission energy to maintain an acceptable bit-error rate.
• Environmental factors such as interference and noise are neglected.

The energy dissipation model has certain limitations:

• The model does not account for dynamic channel variations, hardware imperfections, or environmental interference.
• It assumes uniform node distribution and ideal propagation conditions, which may differ in real deployments.
• Energy consumed by sensing and processing components other than communication is not considered.

3.1. Multilevel Heterogeneity Model

In EATSEP, the node-level diversity is extended to four distinct categories of sensor nodes:

• Normal nodes: Nodes with baseline energy levels lower than all other types.
• Advanced nodes: Nodes with higher energy than normal nodes, but lower than super and ultra-super nodes.
• Super nodes: Nodes with energy levels exceeding those of advanced nodes, but less than ultra-super nodes.
• Ultra-super nodes: Nodes with the highest energy levels in the network, surpassing all other categories.

This design enables a generalized analysis of the impact of increased nodes’ diversity on energy efficiency in wireless sensor networks (WSNs).

The energy allocation for different node categories is defined as follows:

$$E_{ultra-super}=E_{\circ }(1+\alpha )$$

(2)

$$E_{super}=E_{\circ }(1+\beta )$$

(3)

$$E_{adv}=E_{\circ }(1+\gamma )$$

(4)

$$E_{normal}=E_{\circ }$$

(5)

where $E_{\circ }$ is the initial energy of normal nodes, while $\alpha$, $\beta$, and $\gamma$ denote the additional energy factors of ultra-super, super, and advanced nodes, respectively.

The total network energy is expressed as follows:

$$E_{total}=n{E}_{\circ }(1-m-m_{\circ }-n\circ )+nm{E}_{\circ }(1+\alpha )+n{m}_{\circ }{E}_{\circ }(1+\beta )+n{n}_{\circ }{E}_{\circ }(1+\gamma )$$

(6)

Also,

$$E_{normal}=n{E}_{\circ }(1+m\alpha +m_{\circ }\beta +n_{\circ }\gamma )$$

(7)

where n is the total number of nodes, m is the fraction of ultra-super nodes, $m_{\circ }$ is the fraction of super nodes, and $n_{\circ }$ is the fraction of advanced nodes.

Equation (7) shows that EATSEP provides a fraction of $(1+m\alpha +m_{\circ }\beta +n_{\circ }\gamma )$ more energy compared to LEACH [4,5], while maintaining the same number of nodes and deployment area. This highlights its enhanced energy-efficient capability.

3.2. Cluster-Head (CH) Selection

The CH selection in EATSEP follows the same probabilistic approach as in LEACH, but with modified probabilities to account for heterogeneity. The selection probabilities are as follows:

$$P_{normal}={\displaystyle \frac{P_{opt}}{(1+m\alpha +m_{\circ }\beta +n_{\circ }\gamma )}}$$

(8)

$$P_{adv}={\displaystyle \frac{P_{opt}(1+\gamma )}{(1+m\alpha +m_{\circ }\beta +n_{\circ }\gamma )}}$$

(9)

$$P_{super}={\displaystyle \frac{P_{opt}(1+\beta )}{(1+m\alpha +m_{\circ }\beta +n_{\circ }\gamma )}}$$

(10)

$$P_{\mathrm{ultra}-\mathrm{super}}={\displaystyle \frac{P_{opt}(1+\alpha )}{(1+m\alpha +m_{\circ }\beta +n_{\circ }\gamma )}}$$

(11)

The corresponding threshold values for CH election are defined as follows:

$$T_{normal}=\left\{\begin{array}{c}{\displaystyle \frac{P_{normal}}{1-P_{normal}\left[rmod\left(\frac{1}{P_{normal}}\right)\right]}},\mathrm{if}\mathrm{normal}\in {\mathrm{S}}^{\prime \prime \prime \prime }\hfill \\ 0,\mathrm{Otherwise}\hfill \end{array}\right.$$

(12)

$$T_{adv}=\left\{\begin{array}{c}{\displaystyle \frac{P_{adv}}{1-P_{adv}\left[rmod\left(\frac{1}{P_{adv}}\right)\right]}},\mathrm{if}\mathrm{adv}\in {\mathrm{S}}^{\prime \prime \prime }\hfill \\ 0,\mathrm{Otherwise}\hfill \end{array}\right.$$

(13)

$$T_{super}=\left\{\begin{array}{c}{\displaystyle \frac{P_{super}}{1-P_{super}\left[rmod\left(\frac{1}{P_{super}}\right)\right]}},\mathrm{if}\mathrm{super}\in {\mathrm{S}}^{\prime \prime }\hfill \\ 0,\mathrm{Otherwise}\hfill \end{array}\right.$$

(14)

$$T_{\mathrm{ultra}-\mathrm{super}}=\left\{\begin{array}{c}{\displaystyle \frac{P_{\mathrm{ultra}-\mathrm{super}}}{1-P_{\mathrm{ultra}-\mathrm{super}}\left[rmod\left(\frac{1}{P_{\mathrm{ultra}-\mathrm{super}}}\right)\right]}},\mathrm{if}\mathrm{ultra}-\mathrm{super}\in {\mathrm{S}}^{\prime }\hfill \\ 0,\mathrm{Otherwise}\hfill \end{array}\right.$$

(15)

where $S^{\prime }$, $S^{″}$, $S^{‴}$, and $S^{″″}$ represent the sets of ultra-super, super, advanced, and normal nodes respectively that have not been elected as CHs in the current epoch or sub-epoch.

3.3. Hybrid Routing Strategy

To achieve an optimal balance between energy efficiency and data reliability, EATSEP employs a hybrid routing strategy that integrates both reactive and proactive communication mechanisms. This design allows the network to dynamically adapt its behavior based on environmental changes and application requirements. By combining these two modes, the protocol minimizes unnecessary transmissions while ensuring that critical information is always delivered to the base station in a timely manner.

3.3.1. Reactive Routing

In the reactive phase, nodes transmit data only when significant changes occur in the monitored environment. This event-driven behavior reduces redundant transmissions and conserves node energy. EATSEP adopts a dual-threshold mechanism inspired by traditional threshold-sensitive protocols, defined as follows:

• Hard Threshold (HT): An absolute sensed value that triggers a node to switch on its transmitter and report the measurement to its CH.
• Soft Threshold (ST): A minimum change in the sensed attribute that prompts data transmission to the CH, provided the HT has already been reached.

Each node stores the most recent sensed value in an internal variable called SV. Data transmission occurs only if one of the following applies:

• The current sensed value ≥ HT;  
• The difference between the current sensed value and the stored SV ≥ ST.

This mechanism ensures that data is only transmitted when a significant change occurs, thereby conserving energy.

3.3.2. Proactive Routing

If no attribute surpasses the HT during a given interval, CHs transmit data to the base station after a pre-defined number of rounds. This periodic reporting ensures continuous monitoring, providing reliability for safety-critical applications, albeit at the cost of additional energy consumption. The increased energy heterogeneity of EATSEP compensates for this overhead.

The overall procedure of the proposed scheme is presented in Algorithm 1, and a corresponding flow chart is provided in Figure 2 for additional clarity.

Algorithm 1: Proposed Protocol.

3.4. Numerical Explanation

Let us consider a sensor network of 100 nodes randomly deployed over a field of size $M\times M{\mathrm{m}}^2$. The network is heterogeneous, with four levels of nodes. In this configuration, $20\%$ of the nodes ($n_{\circ }=0.2$) are advanced nodes, each endowed with $300\%$ more energy than normal nodes ($\gamma =3$). Similarly, $10\%$ of the nodes ($m_{\circ }=0.1$) are super nodes with $200\%$ more energy than normal nodes ($\beta =2$), and another $10\%$ of the nodes ($m=0.1$) are ultra-super nodes, also endowed with $200\%$ more energy than normal nodes ($\alpha =2$). The remaining $60\%$ of the network is composed of normal nodes with baseline energy levels.

The cluster-head (CH) selection probability for each class of nodes is adjusted according to its relative energy level. With the optimal probability of cluster-head selection set as $P_{\mathrm{opt}}=0.1$, the denominator factor becomes $D=1+\alpha m+\beta m_{\circ }+\gamma n_{\circ }=2.0$. Substituting this into the probability expressions, we obtain $P_{\mathrm{normal}}=0.05$, $P_{\mathrm{adv}}=0.20$, $P_{\mathrm{super}}=0.15$, $P_{\mathrm{ultra}-\mathrm{super}}=0.15$. Given that the network has 60 normal nodes, 20 advanced nodes, 10 super nodes, and 10 ultra-super nodes, the expected number of cluster-heads per round is approximately 3 from the normal nodes, 4 from the advanced nodes, $1.5$ from the super nodes, and $1.5$ from the ultra-super nodes. This adds up to 10 cluster-heads per round, which matches the expected outcome of $n\times P_{\mathrm{opt}}=100\times 0.1$.

In the homogeneous case of LEACH, an epoch consists of ${\displaystyle \frac{1}{P_{\mathrm{opt}}}}=10\mathrm{rounds}$. However, in this heterogeneous setting, the extended epoch length is calculated as

$${\displaystyle \frac{1+\alpha m+\beta m_{\circ }+\gamma n_{\circ }}{P_{\mathrm{opt}}}}=20\mathrm{rounds}$$

(16)

Within this epoch, each node type follows its own sub-epoch cycle. Advanced nodes, for instance, complete their cycle in ${\displaystyle \frac{1}{P_{\mathrm{opt}}}}·{\displaystyle \frac{D}{1+\gamma }}={\displaystyle \frac{10\times 2}{4}}=5\mathrm{rounds}$. Super nodes and ultra-super nodes each require approximately ${\displaystyle \frac{1}{P_{\mathrm{opt}}}}·{\displaystyle \frac{D}{1+\beta }}={\displaystyle \frac{10\times 2}{3}}\approx 6.67\mathrm{rounds}$, ${\displaystyle \frac{1}{P_{\mathrm{opt}}}}·{\displaystyle \frac{D}{1+\alpha }}={\displaystyle \frac{10\times 2}{3}}\approx 6.67\mathrm{rounds},$ respectively. Normal nodes, by contrast, span the entire 20-round heterogeneous epoch.

This ensures that within each epoch or sub-epoch, nodes of every class are elected as cluster-heads exactly once on average, without repetition during their assigned cycle. For instance, each of the 60 normal nodes becomes a cluster-head once in the 20-round heterogeneous epoch, while each advanced node becomes a cluster-head once within 5 rounds. Similarly, each super node and each ultra-super node is elected as a cluster-head once within approximately 7 rounds. In this way, the cluster-head selection is fairly distributed across all categories of nodes according to their energy levels. This balanced and energy-aware mechanism demonstrates the protocol’s capability to exploit multilevel energy variation, ensuring both fairness and energy efficiency across the entire network.

The contributions of EATSEP can be summarized as follows:

• An enhanced multilevel heterogeneity model (normal, advanced, super, ultra-super) for efficient energy utilization.
• A probability-based CH election mechanism adapted to heterogeneous energy levels.
• A hybrid routing approach integrating reactive and proactive communication to balance energy efficiency and reliability.

3.5. Novelty and Distinctive Features of the Proposed Model

Unlike previous enhancements to TSEP and SEP that primarily focus on optimizing cluster-head (CH) selection through residual energy, distance, or zonal division, the proposed model introduces a multilevel heterogeneity-based architecture that integrates both energy-level stratification and adaptive CH selection across unequal energy tiers. Novelty and distinctive features of the proposed model are as follows:

• Extended Heterogeneity: Prior works such as EATSEP [18] and ITSEP [19] consider one or two heterogeneity levels (normal and advanced nodes). In comparison, our model generalizes this concept to multiple (three or more) energy levels, capturing finer-grained energy diversity among nodes.
• Adaptive Hybrid Routing Extension: The proposed model enhances this strategy by coupling it with multilevel heterogeneity awareness. Cluster-heads dynamically adjust the frequency of proactive reporting and event thresholds based on their residual energy level and heterogeneity tier, ensuring that higher-energy nodes handle more frequent transmissions while conserving the energy of lower-tier nodes. This adaptive coupling between routing behavior and node energy levels further improves both energy efficiency and data reliability, outperforming static hybrid schemes.
• Inter-Level Energy Awareness: Rather than relying on spatial or zonal partitioning as in HISEP [20], IZ-SEP [21], and ZET [22], the proposed model establishes energy-aware coordination between heterogeneity levels via its hybrid routing design. Nodes with higher residual energy predominantly engage in proactive transmissions, ensuring regular data delivery and network reliability, while lower-energy nodes operate mainly in reactive mode, transmitting only upon significant environmental changes. This inter-level collaboration enables high-energy nodes to effectively support low-energy ones during critical phases, balancing overall energy consumption and enhancing network longevity.
• Performance under Unequal Energy Variations: The model is evaluated under unequal and realistic energy distributions, demonstrating robustness and scalability beyond the assumptions of equal zonal densities used in prior protocols.

4. Simulation Results

Simulation results were generated using MATLAB R2016a software to evaluate the performance of the proposed EATSEP protocol against existing benchmark protocols, namely LEACH [4,5], SEP [11], and TSEP [13]. A wireless sensor network consisting of 100 nodes randomly deployed within a square field of size $(M\times M){\mathrm{m}}^2$ was considered. The base station (BS) or sink was assumed to be located at the center of the sensing field.

The optimal probability of cluster-head (CH) selection was set as $P_{opt}=0.1$, meaning that, on average, $10\%$ of the nodes act as CHs in each round. Based on this value, the CH selection probabilities for normal, advanced, super, and ultra-super nodes were computed using Equations (8)–(11). Their corresponding threshold values for CH election were determined according to Equations (12)–(15).

The detailed simulation parameters employed in this study are summarized in

Table 1. Simulation parameters.

Parameters	Values
$E_{elect}$(Energy dissipated per bit to run the transmitter or the receiver circuit)	50 nJ/bit
$E_{DA}$ (Processing cost of a bit per report to the sink)	5 nJ/bit/message
${ϵ}_{f_s}$ (Amplification energy when $d<d_{\circ }$)	10 pJ/bit/${\mathrm{m}}^2$
${ϵ}_{mp}$ (Amplification energy when $d\ge d_{\circ }$)	0.01 3pJ/bit/${\mathrm{m}}^4$
$E_{\circ }$ (Initial energy of the normal nodes)	0.5 J
Message size	4000 bits
$P_{opt}$ (Optimal probability)	0.1
N (Total number of nodes in the field area)	100 nodes
Network field	100 m × 100 m
$d_{\circ }$ (Threshold distance)	87.706 m

.

4.1. Case 1: $\alpha$ = 3, $\beta$ = $\alpha$/2, $\gamma$ = $\alpha$/4, m = 0.2, $m_{\circ }$ = 0.2, $n_{\circ }$ = 0.3

Figure 3 presents a comparative analysis of LEACH [4,5], SEP [11], TSEP [13], and the proposed EATSEP protocols in terms of the number of dead nodes versus the number of rounds. The simulation parameters were configured with $\alpha =3$, $\beta ={\displaystyle \frac{\alpha }{2}}$, $\gamma ={\displaystyle \frac{\alpha }{4}}$, $m=0.2$, $m_{\circ }=0.2$ and $n_{\circ }=0.3$.

The results clearly indicate that LEACH, being a homogeneous protocol, and SEP, which introduces only two levels of heterogeneity, achieve the lowest stability periods. Here, the stability period is defined as the duration from the beginning of network operation until the death of the first node. The shorter stability phase of these protocols reflects their lower energy efficiency and reduced reliability compared to the others.

In contrast, TSEP, which incorporates three levels of heterogeneity and therefore provides additional energy to selected nodes, demonstrates improved performance. It achieves a longer system stability duration than both LEACH and SEP, highlighting the benefits of introducing non-uniformity among nodes’ energy distribution.

We further introduce two versions of the proposed EATSEP protocol, designed with three $(H=3)$ and four $(H=4)$ levels of heterogeneity, respectively. Both versions outperform the baseline protocols in terms of stability and overall network lifetime. Among them, EATSEP $(H=4)$ demonstrates superior results by achieving longer stability and instability periods compared to EAPTSEP $(H=3)$. This clearly illustrates that increasing the degree of nodes’ diversity not only enhances energy efficiency but also extends network lifetime, thereby supporting diverse application requirements and improving the adaptability of the system to varying network demands.

Figure 4 illustrates the comparative performance of LEACH [4,5], SEP [11], TSEP [13], and the proposed EATSEP protocol in terms of the number of packets transmitted from cluster-heads (CHs) to the base station (BS) over successive rounds. The results clearly demonstrate that EATSEP achieves the highest throughput among all protocols. This superior performance is attributed to its uneven nodes’ energy distribution and its hybrid routing mechanism, which together enable both efficient energy utilization and frequent data reporting. Following EATSEP, TSEP records the second-highest number of packet transmissions from CHs to BS, benefiting from three levels of heterogeneity that provide nodes with additional energy reserves.

In contrast, SEP (two levels of heterogeneity) and LEACH (homogeneous) achieve the lowest throughput. While both protocols allow continuous data transmission, the limited energy available in their nodes restricts the total number of packets delivered to the BS. Consequently, they rank fourth and fifth in the comparative analysis, respectively.

For case 1, a comparison of LEACH (H = 1) [4,5], SEP (H = 2) [11], TSEP (H = 3) [13], EATSEP (H = 3), and EATSEP (H = 4) is presented in

Table 2. Comparison of LEACH (H = 1), SEP (H = 2), TSEP (H = 3), EATSEP (H = 3) and EATSEP (H = 4) based on simulation results (Case 1).

Protocols	Death of 1st Node (Rounds)	Iteration of Node Death (Rounds)	Data Sent from CHs to BS (Packets)
LEACH (H = 1) [4,5]	783	All nodes died between 2000 and 3000 rounds	48,070
SEP (H = 2) [11]	1240 (+58% vs. LEACH, H = 1)	All nodes died between 4000 and 5000 rounds	61,480
TSEP (H = 3) [13]	1567 (+26% vs. SEP, H = 2)	All nodes died between 5000 and 6000 rounds	43,560
EATSEP (H = 3)	2492 (+59% vs. TSEP, H = 3)	All nodes died between 7000 and 8000 rounds	70,390
EATSEP (H = 4)	3146 (+26% vs. EATSEP, H = 3)	All nodes died between 12,000 and 13,000 rounds	153,600

.

4.2. Case 2: $\alpha$ = 4, $\beta$ = $\alpha$/2, $\gamma$ = $\alpha$/4, m = 0.3, $m_{\circ }$ = 0.2, $n_{\circ }$ = 0.3

Figure 5 shows the simulation results comparing LEACH [4,5], SEP [11], TSEP [13], and the proposed EATSEP protocol, focusing on the relationship between the number of dead nodes and the number of rounds. This evaluation follows the same methodology as in Case 1, but with modified parameters—$\alpha =4$ and $m=0.3$—while all other parameters remain unchanged. These adjustments allow us to examine the impact of increasing both the additional energy factor and the fraction of ultra-super nodes on network performance.

With the higher values of $\alpha$ and m, the energy reserves of the nodes increase significantly, along with a corresponding increase in the probability of cluster-head (CH) selection. This configuration enables the network to sustain operations for a greater number of rounds, as observed in Figure 5. The results clearly show that the additional network energy not only extends the overall network lifetime but also enhances stability, delaying the death of the first node and thereby prolonging the steady-state period.

Figure 6 presents the simulation results comparing LEACH [4,5], SEP [11], TSEP [13], and the proposed EATSEP in terms of packets transmitted from cluster-heads (CHs) to the base station (BS) over successive rounds with updated parameters ($\alpha =4$ and $m=0.3$) to assess the effect of higher heterogeneity on network throughput.

With the higher values of $\alpha$ and m, nodes are endowed with greater energy capacity and the probability of cluster-head selection rises accordingly. As a result, the network is able to sustain a higher volume of data transmissions over a longer duration. This leads to a noticeable increase in throughput, as depicted in Figure 6.

The results confirm that increasing the energy disparity among nodes directly enhances its ability to deliver packets from CHs to the BS. The additional energy reserves not only extend network lifetime but also enable more consistent and higher data delivery rates, reinforcing the advantage of multilevel heterogeneous designs such as EATSEP over traditional baseline protocols.

For case 2, a comparison of LEACH (H = 1) [4,5], SEP (H = 2) [11], TSEP (H = 3) [13], EATSEP (H = 3), and EATSEP (H = 4) is presented in

Table 3. Comparison of LEACH (H = 1), SEP (H = 2), TSEP (H = 3), EATSEP (H = 3) and EATSEP (H = 4) based on simulation results (Case 2).

Protocols	Death of 1st Node (Rounds)	Iteration of Node Death (Rounds)	Data Sent from CHs to BS (Packets)
LEACH (H = 1) [4,5]	819	All nodes died between 1000 and 2000 rounds	47,970
SEP (H = 2) [11]	1386 (+69% vs. LEACH, H = 1)	All nodes died between 6000 and 7000 rounds	68,200
TSEP (H = 3) [13]	1636 (+18% vs. SEP, H = 2)	All nodes died between 5000 and 6000 rounds	46,040
EATSEP (H = 3)	2612 (+60% vs. TSEP, H = 3)	All nodes died between 8000 and 9000 rounds	74,550
EATSEP (H = 4)	2977 (+14% vs. EATSEP, H = 3)	All nodes died between 17,000 and 18,000 rounds	162,400

.

4.3. Case 3: $\alpha$ = 5, $\beta$ = $\alpha$/2, $\gamma$ = $\alpha$/4, m = 0.4, $m_{\circ }$ = 0.2, $n_{\circ }$ = 0.3

In Case 3, the simulation was conducted with $\alpha =4$ and $m=0.4$, while all other parameters remained unchanged. This scenario further increases the additional energy factor and the fraction of ultra-super nodes in the network to investigate the effect of energy diversity on network performance.

The simulation results were consistent with those observed in Case 1 ($\alpha =4$ and $m=0.4$), showing the same trends in stability, node survival, and throughput. However, with the increased energy provision to ultra-super nodes, the network demonstrated longer overall lifetime and a higher data delivery ratio as shown in Figure 7 and Figure 8.

These results confirm that enhancing the non-uniformity by increasing both the energy factor and the proportion of high-energy nodes provides significant performance gains. The network can sustain operations for a greater number of rounds while maintaining reliable data transmission from cluster-heads to the base station, illustrating the effectiveness of EATSEP in supporting energy-efficient and robust wireless sensor networks.

For case 3, a comparison of LEACH (H = 1) [4,5], SEP (H = 2) [11], TSEP (H = 3) [13], EATSEP (H = 3), and EATSEP (H = 4) is presented in

Table 4. Comparison of LEACH (H = 1), SEP (H = 2), TSEP (H = 3), EATSEP (H = 3) and EATSEP (H = 4) based on simulation results (Case 3).

Protocols	Death of 1st Node (Rounds)	Iteration of Node Death (Rounds)	Data Sent from CHs to BS (Packets)
LEACH (H = 1) [4,5]	762	All nodes died between 1000 and 2000 rounds	48,060
SEP (H = 2) [11]	1473 (+93% vs. LEACH, H = 1)	All nodes died between 5000 and 6000 rounds	70,130
TSEP (H = 3) [13]	1675 (+14% vs. SEP, H = 2)	All nodes died between 10,000 and 11,000 rounds	50,050
EATSEP (H = 3)	2831 (+69% vs. TSEP, H = 3)	All nodes died between 14,000 and 15,000 rounds	79,800
EATSEP (H = 4)	3087 (+9% vs. EATSEP, H = 3)	All nodes died after more than 18,000 rounds	177,300 (+) 1

¹ Note: “(+)” denotes that the value is greater than the number shown.

.

The MATLAB simulations provide several key insights into the performance of the proposed EATSEP protocol:

• Stability and Network Lifetime: EATSEP demonstrates higher stability and a longer network lifetime compared to LEACH [4,5], SEP [11], and TSEP [13].
• Comparison with TSEP: TSEP [13] is less suitable for periodic data transmission and continuous network updates, which are critical for safety-critical applications. Therefore, EATSEP is better suited for such scenarios.
• Throughput: EATSEP achieves the highest throughput among all compared protocols, benefiting from both its hybrid routing strategy and the four levels of heterogeneity that enhance energy utilization.
• Effect of Heterogeneity Levels: Among the protocols under consideration, LEACH [4,5] is homogeneous, SEP [11] is heterogeneous with two levels, TSEP [13] employs three levels, and EATSEP incorporates multilevel energy diversity. Examining Figure 3, Figure 5 and Figure 7, it is evident that increasing the heterogeneity levels directly improves network reliability and energy efficiency. When nodes follow the proposed CH selection strategy and efficiently utilize the extra energy made available by increased non-uniformity, the overall network performance is significantly enhanced.

In summary, the simulation results confirm that the proposed EATSEP protocol leverages multilevel heterogeneity and hybrid routing to achieve superior stability, energy efficiency, and throughput, making it highly suitable for modern wireless sensor networks with diverse operational requirements.

5. Discussion

The performance of the proposed multilevel heterogeneity-based WSN model was evaluated through MATLAB simulations and compared with existing clustering protocols, including LEACH (H = 1), SEP (H = 2), TSEP (H = 3), and EATSEP (H = 3, 4). Key performance indicators considered were network lifetime, steady-state period, throughput, and energy efficiency.

Simulation results revealed that the proposed model achieved a substantial improvement in network lifetime compared to all benchmark protocols. As the level of energy diversity increased, both the first node death and last node death were delayed, indicating better energy utilization and load distribution. The model with four levels of heterogeneity (H = 4) demonstrated the longest early-phase network lifetime, maintaining active communication for a significantly extended duration before the first node failure.

Throughput analysis showed a consistent rise with increasing node-level diversity, primarily due to the optimized cluster-head (CH) selection mechanism that reduces redundant communication and balances node energy consumption. The proposed model achieved higher packet delivery rates and minimized energy depletion across the network compared to LEACH, SEP, and TSEP.

From the statistical analysis, the proposed architecture exhibited a best-case improvement of approximately 9000 additional rounds and a worst-case improvement of about 3000 rounds when transitioning from three to four heterogeneity levels. This significant performance gain highlights the impact of introducing an additional heterogeneity layer. Moreover, the findings confirm that not only the level of energy diversity but also the distribution and quantity of nodes in each level critically influence network performance and longevity. We deployed 100 nodes randomly across the sensing field. Increasing node density in a WSN generally improves coverage, reliability, and fault tolerance, as more nodes can detect events and compensate for failures. It can also extend network lifetime if clustering is efficient, since nodes can share the load of cluster-head duties. However, higher density may increase energy consumption, communication collisions, and routing load, especially if the protocol does not manage cluster formation and data transmission efficiently.

In the proposed model, the hybrid reactive–proactive routing helps mitigate these issues by adapting transmissions based on node energy levels and event significance, maintaining efficiency even as node density rises. In reactive mode, nodes transmit only upon detecting significant environmental changes, ensuring timely reporting of critical events such as node failures or hazards. In proactive mode, periodic transmissions from cluster-heads maintain continuous network awareness, even under unstable conditions like variable base station positions or partial node losses.

In short, the observed enhancements in lifetime, throughput, and energy efficiency demonstrate that introducing unequal energy distribution effectively balances network load and mitigates premature node deaths. The optimized CH selection algorithm ensures that high-energy nodes are utilized efficiently without overburdening them. Consequently, the proposed system’s reactive–proactive routing achieves a more stable and energy-balanced operation, making it particularly suitable for disaster management, environmental monitoring, and other safety-critical WSN applications.

6. Conclusions

In this work, we have proposed EATSEP, a multilevel heterogeneous, hybrid routing protocol designed to address the requirements of safety-critical applications in wireless sensor networks (WSNs). Unlike TSEP, which relies solely on a threshold-sensitive mechanism, EATSEP supports both time-critical and periodic data transmission, making it better suited for practical applications such as disaster management, early warning systems, and environmental monitoring. Comparative analysis with LEACH, SEP, and TSEP demonstrates that EATSEP consistently outperforms these protocols in terms of network lifetime, throughput, and energy efficiency, particularly due to its multilevel heterogeneity and optimized cluster-head selection strategy.

The results further confirm that increasing heterogeneity in sensor networks significantly enhances reliability and energy efficiency when nodes follow the proposed CH selection and energy utilization scheme. Overall, the proposed EATSEP protocol provides a robust, energy-efficient, and reliable framework for WSNs, making it highly suitable for safety-critical and mission-critical applications where both timely and continuous monitoring are required. Although real-world deployment was not conducted, simulation results demonstrate that this dual-mode routing enhances adaptability and reliability under challenging scenarios. Future work will extend this approach to include node and base station mobility, dynamic clustering mechanisms to further improve adaptability in rapidly changing environments, security integration, and heterogeneous sensing, enabling robust performance in realistic disaster management systems.