Hyper-heuristic driven sustainable wireless networks for crisis-resilient communication systems

1 Introduction

Reliable communication infrastructures play a critical role in supporting emergency response and disaster management operations. During large-scale crises such as earthquakes, floods, wildfires, and armed conflicts, conventional communication infrastructure is often partially or completely destroyed. In such circumstances, rapidly deployable wireless networking technologies become essential for enabling coordination among rescue teams, medical services, humanitarian organizations, and affected populations. Recent comprehensive surveys have highlighted the critical need for sustainable and resilient communication infrastructure solutions capable of operating effectively under diverse disaster relief and management scenarios (Karaman et al., 2025).

Wireless communication paradigms such as Mobile Ad Hoc Networks (MANETs), Wireless Sensor Networks (WSNs), and UAV-assisted wireless networks have been widely proposed as flexible communication solutions for disaster response scenarios. These networks can be deployed quickly without relying on pre-existing infrastructure and are capable of providing temporary connectivity in damaged environments. In particular, UAV swarm networks have attracted significant research attention for disaster emergency communications, where spatiotemporal optimization approaches based on deep reinforcement learning have been proposed to coordinate UAV positioning and enhance communication coverage in affected regions (Mei et al., 2026). However, crisis environments introduce several significant challenges that severely affect the performance and reliability of wireless communication systems.

First, the network topology in disaster environments is highly dynamic due to node mobility, device failures, and unpredictable environmental conditions. Second, wireless nodes are typically powered by limited battery resources, making energy efficiency a critical requirement for sustaining communication services over extended periods. Third, emergency communication networks often experience irregular traffic patterns caused by sudden bursts of critical information exchange. Finally, large-scale node failures and intermittent connectivity may lead to frequent route disruptions and reduced network reliability.

Traditional routing and optimization strategies are generally designed for relatively stable network environments and therefore struggle to cope with the extreme dynamics and uncertainties encountered in crisis situations. Static routing protocols and single-heuristic optimization approaches lack the flexibility required to adapt to rapidly changing network conditions. Recent studies have explored deep reinforcement learning-based resilient routing strategies for urban emergency communication networks, demonstrating the potential of intelligent adaptive mechanisms to maintain network performance under disrupted conditions (Jin et al., 2025). Furthermore, computerized intelligent systems have been developed for optimizing UAV communication relay planning in emergency rescue operations, emphasizing the importance of automated decision-making frameworks in time-critical disaster response scenarios (Ni et al., 2026). Consequently, there is a strong need for adaptive and intelligent decision-making mechanisms capable of dynamically selecting the most appropriate communication strategies based on the current network state.

Recent advances in hyper-heuristic optimization offer promising opportunities to address these challenges. Hyper-heuristics operate at a higher abstraction level compared with traditional metaheuristics by focusing on the automated selection or generation of heuristics rather than directly solving the optimization problem itself. By maintaining a pool of candidate low-level heuristics and selecting among them dynamically, hyper-heuristics provide a flexible framework capable of adapting optimization strategies to evolving system conditions.

Motivated by these advantages, this paper proposes a hyper-heuristic driven framework designed to improve the sustainability, adaptability, and resilience of wireless communication networks deployed in crisis environments. The proposed approach integrates multiple communication heuristics categorized into constructive, improvement, perturbation, and reconstructive strategies. A high-level hyper-heuristic controller dynamically selects the most suitable heuristic based on real-time network performance indicators such as residual energy levels, link quality, congestion conditions, and network density.

In addition, the framework incorporates a multi-objective optimization model that simultaneously considers energy consumption, communication delay, packet loss rate, and network reliability. This formulation enables the system to balance multiple performance objectives while maintaining sustainable network operation in highly constrained disaster scenarios.

Extensive simulation experiments conducted using the NS-3 network simulator demonstrate that the proposed hyper-heuristic framework significantly improves network lifetime, packet delivery performance, and energy efficiency compared with conventional routing strategies.

2 Related work

Wireless communication technologies have become a critical component in disaster management and emergency response systems (Hu et al., 2026; Baraneedharan et al., 2026; Abidde et al., 2025; Chandran and Vipin, 2024). When conventional communication infrastructure becomes unavailable due to natural disasters, conflicts, or large-scale emergencies, rapidly deployable wireless networks such as Mobile Ad Hoc Networks (MANETs), Wireless Sensor Networks (WSNs), and UAV-assisted networks can provide temporary connectivity and support coordination among rescue teams and affected populations (Çoğay et al., 2026; Hydher et al., 2025). In parallel, intelligent optimization methods such as hyper-heuristics have gained attention as adaptive mechanisms for managing complex and dynamic systems (Wang et al., 2025). This section reviews previous research in three main areas: crisis communication networks, energy-efficient wireless networking, and hyper-heuristic optimization techniques.

2.1 Wireless networks for crisis communication

Several studies have investigated the use of infrastructure-less wireless networks for disaster response scenarios. Wireless mesh networks and ad hoc networks can be rapidly deployed in affected areas and provide flexible communication services under highly dynamic conditions (Abidde et al., 2025). Early surveys on wireless networking technologies highlighted their potential for supporting emergency communication services in infrastructure failure situations (Akyildiz et al., 2005; Li, 2024).

More recently, UAV-assisted wireless networks have been proposed to extend network coverage and restore connectivity in disaster regions (Chandran and Vipin, 2024; Biswas et al., 2026). UAV relays can dynamically reposition themselves to bridge disconnected network segments and provide aerial communication support (Mozaffari et al., 2019; Çoğay et al., 2026). These approaches significantly improve network coverage and resilience, particularly in geographically complex disaster environments (Khan et al., 2024).

Despite these advancements, crisis communication networks still suffer from several operational challenges including unstable topology, intermittent connectivity, and limited energy resources (Hydher et al., 2025). Efficient routing and adaptive network management strategies remain essential to ensure reliable communication in such environments (Liu et al., 2025).

2.2 Energy-efficient wireless networking

Energy efficiency is one of the most critical challenges in wireless networking, especially in disaster scenarios where nodes typically operate on limited battery power (Liu et al., 2019). Numerous routing protocols have been proposed to minimize energy consumption and extend network lifetime (Abbasi and Hadi, 2023).

Energy-aware routing protocols attempt to balance energy consumption across nodes by selecting communication paths based on residual energy levels and transmission costs (Heinzelman et al., 2000; Karaki and Kamal, 2004). Cluster-based communication techniques have also been widely investigated to improve scalability and reduce communication overhead in wireless sensor networks (Younis and Fahmy, 2004; Abbasi and Younis, 2007).

Recent research has further explored energy harvesting and sustainable communication mechanisms to prolong network lifetime (Shaikh and Pathan, 2016). However, many existing approaches rely on static optimization strategies that may not adapt effectively to rapidly changing network conditions typically encountered during emergency operations (Kuhnen, 2021).

2.3 Hyper-heuristic optimization

Hyper-heuristics have emerged as a powerful class of high-level optimization methods designed to automate the selection or generation of heuristics for solving complex optimization problems (Danach et al., 2025a). Unlike traditional metaheuristics that directly explore the solution space, hyper-heuristics operate on a space of heuristics and aim to identify the most suitable strategy for a given problem instance (Burke et al., 2013).

The concept of hyper-heuristics was extensively investigated in combinatorial optimization, where heuristic selection and generation approaches were developed to improve algorithm adaptability and generality (Burke et al., 2010; Zhong et al., 2019; Danach et al., 2025b). Recent surveys have emphasized the potential of hyper-heuristics in dynamic optimization environments where problem characteristics evolve over time (Ochoa and Drake, 2019).

Although hyper-heuristic approaches have been successfully applied in domains such as scheduling (Luo et al,. 2026), logistics (Danach et al., 2026), and resource allocation (Chao et al., 2026), their application to wireless networking problems remains relatively limited. In particular, few studies have explored the integration of hyper-heuristic decision mechanisms for adaptive routing and resource management in crisis communication networks.

2.4 Research gap

From the reviewed literature, it can be observed that significant progress has been made in both wireless disaster communication systems and hyper-heuristic optimization techniques. Recent studies have investigated UAV-assisted emergency communication, resilient routing protocols, machine learning-based optimization, and sustainable wireless networking frameworks for disaster-response operations. However, most existing wireless networking approaches rely on static routing strategies, predefined optimization policies, or single meta-heuristic mechanisms that may not perform effectively under highly dynamic crisis conditions.

Despite the considerable advancements achieved in resilient wireless communication research, several important limitations remain insufficiently addressed. Existing approaches often provide limited adaptability to continuously changing network conditions such as node mobility, congestion variation, dynamic topology changes, link instability, and energy depletion. Furthermore, many optimization frameworks focus primarily on isolated objectives such as energy efficiency or delay reduction without simultaneously considering sustainability, communication reliability, and adaptive decision-making in integrated disaster communication environments.

Furthermore, while hyper-heuristics provide powerful adaptive optimization capabilities, their application to sustainable wireless networking in disaster environments has received limited attention. Limited research has investigated adaptive hyper-heuristic frameworks capable of dynamically selecting communication heuristics according to real-time network states and operational constraints.

There is therefore a need for intelligent frameworks capable of dynamically selecting communication strategies according to real-time network conditions while simultaneously optimizing energy efficiency and communication reliability.

To address this gap, this paper proposes a hyper-heuristic driven framework for sustainable wireless networks designed specifically for crisis-resilient communication systems. The proposed framework integrates adaptive heuristic selection, real-time network state evaluation, and multi-objective optimization in order to enhance network sustainability, resilience, and communication performance in highly dynamic disaster-response environments.

3 System model

This section presents the system model adopted for the proposed crisis-resilient wireless networking framework. The model describes the network architecture, communication assumptions, and the energy consumption characteristics of the deployed nodes.

3.1 Network architecture

We consider a rapidly deployable wireless communication system designed for emergency and disaster scenarios where conventional communication infrastructure is partially or completely unavailable. The network is modeled as a distributed and infrastructure-less wireless system composed of heterogeneous nodes deployed across the affected area.

Let the network be represented by a graphwhere V = { v 1 , v 2 , … , v N } denotes the set of wireless nodes and E represents the set of wireless communication links between nodes. Each edge ( v i , v j ) ∈ E indicates that nodes v i and v j are within communication range and can directly exchange packets.

G = V , E

The network includes three types of nodes:

• Mobile user nodes: handheld devices carried by rescue teams, emergency responders, or civilians.
• Sensor nodes: lightweight sensing devices deployed for environmental monitoring such as structural damage, temperature, smoke detection, or human presence.
• UAV relay nodes: unmanned aerial vehicles acting as aerial communication relays to extend network coverage and restore connectivity between disconnected ground segments.

The deployment is assumed to occur in a geographically constrained disaster region denoted by an area A. Node distribution within the region follows a random spatial distribution that reflects the unpredictable nature of crisis deployments.

Due to node mobility and possible device failures, the network topology evolves dynamically over time. Consequently, maintaining reliable connectivity and efficient routing becomes a critical challenge in such environments.

Figure 1 illustrates the overall architecture of the proposed adaptive crisis communication framework. The system integrates wireless multi-hop communication, UAV-assisted relay support, mobile rescue nodes, environmental sensing infrastructure, and a hyper-heuristic decision controller capable of dynamically selecting routing and communication heuristics according to real-time network conditions. The architecture also demonstrates the interaction between the state evaluation module, heuristic selection mechanism, and optimization objectives for maintaining reliable and energy-efficient communication in disaster environments.

3.2 Communication model

Communication between nodes occurs through wireless multi-hop transmission. Each node is equipped with a single wireless interface operating within a fixed transmission range R. Two nodes can establish a direct communication link if the Euclidean distance between them satisfieswhere d i j denotes the distance between nodes v i and v j.

d i j ≤ R

The wireless channel is assumed to follow a shared medium model where concurrent transmissions may lead to interference and packet collisions. To mitigate this effect, a medium access control mechanism is assumed to regulate channel access.

In practical disaster environments, wireless communication performance may also be affected by environmental noise and channel disturbances caused by damaged infrastructure, electromagnetic interference, harsh environmental conditions, or overlapping wireless transmissions. Such factors may degrade signal quality, increase packet retransmissions, and reduce routing stability. Although explicit noise-aware optimization is beyond the scope of the current study, integrating noise modeling and interference-aware communication mechanisms represents an important direction for future research in crisis-resilient wireless networks.

Data transmission in the network follows a store-and-forward multi-hop routing paradigm. Packets generated by a source node are forwarded through intermediate relay nodes until they reach their intended destination. Due to the highly dynamic topology, routes may frequently change, requiring adaptive routing strategies.

The network state at time t is characterized by a set of observable parameters including:

• Node residual energy
• Link quality indicators
• Queue length and traffic load
• Node mobility patterns
• Network connectivity degree

These parameters are continuously monitored and provided to the hyper-heuristic decision layer responsible for selecting the most appropriate routing and communication strategy.

3.3 Energy model

Energy efficiency is a critical requirement in crisis communication networks since nodes are typically powered by limited battery resources and recharging opportunities are often unavailable.

Each node v i is initialized with a finite energy budget E i init. The residual energy at time t is denoted by E i ( t ).

Energy consumption is modeled according to the widely adopted radio energy model where energy expenditure occurs during three primary operations:

• Packet transmission
• Packet reception
• Idle listening

The energy required to transmit a packet of size l bits over a distance d is given bywhere

E t x l , d = l E elec + l ϵ amp d α

• E elec represents the electronic circuitry energy consumption per bit,
• ϵ amp denotes the transmit amplifier coefficient,
• α is the path-loss exponent typically ranging between 2 and 4 depending on the propagation environment.

The energy consumed during packet reception is defined as

E r x l = l E elec

Additionally, nodes consume energy during idle channel monitoring, which is modeled aswhere P idle represents the idle power consumption and T idle denotes the duration of idle listening.

E idle = P idle × T idle

The residual energy of node v i evolves over time according to

E i t + 1 = E i t − E t x + E r x + E idle

The network lifetime is defined as the time until the first critical node exhausts its energy or when network connectivity drops below a predefined threshold.

Efficient energy utilization therefore becomes essential to prolong the operational lifetime of the network and ensure sustainable communication services during prolonged crisis situations.

The proposed hyper-heuristic framework leverages the energy state of nodes as a key parameter when selecting appropriate communication heuristics, enabling adaptive strategies that balance energy consumption with communication performance.

4 Hyper-heuristic framework

The proposed framework adopts a high-level hyper-heuristic strategy designed to autonomously manage heuristic selection in highly dynamic crisis communication environments. Unlike traditional optimization methods that rely on a single predefined heuristic, the hyper-heuristic operates at a supervisory level and dynamically selects or adapts low-level heuristics according to the observed network state.

The framework consists of three main components: a heuristic pool containing multiple candidate low-level heuristics, a hyper-heuristic controller responsible for heuristic selection and adaptation, and a state evaluation mechanism that continuously monitors network conditions. In addition, the framework incorporates acceptance criteria and termination criteria to guide the search process toward efficient network configurations.

4.1 Heuristic pool

The heuristic pool contains a diverse set of low-level heuristics designed to address different networking objectives such as energy efficiency, latency reduction, load balancing, and connectivity maintenance. In the proposed framework, these heuristics are categorized into four classes commonly used in hyper-heuristic optimization: constructive heuristics, improvement heuristics, perturbation heuristics, and reconstructive heuristics.

Constructive heuristics are responsible for generating initial feasible communication structures or routing configurations. In the context of crisis wireless networks, constructive heuristics establish initial multi-hop routing paths based on criteria such as minimum energy consumption or shortest communication distance.

Improvement heuristics aim to refine an existing routing configuration by making incremental adjustments that enhance performance metrics. Examples include delay-aware routing adjustments and load-balancing mechanisms that redistribute traffic across alternative paths to reduce congestion.

Perturbation heuristics introduce controlled modifications to the current network configuration in order to escape local optima and explore alternative communication structures. Opportunistic forwarding strategies fall within this category, allowing nodes to dynamically select alternative relays when link quality degrades.

Reconstructive heuristics partially or completely rebuild routing structures when the current configuration becomes inefficient due to topology changes or node failures. Cluster-based communication mechanisms can be applied in this phase to reorganize the network into energy-efficient clusters and improve scalability.

The candidate heuristics integrated into the pool include:

• Energy-aware routing
• Delay-aware routing
• Load-balanced routing
• Cluster-based communication
• Opportunistic forwarding

By maintaining a heterogeneous set of heuristics, the system can adapt to different operational requirements encountered during crisis scenarios.

4.2 Hyper-heuristic controller

The hyper-heuristic controller acts as the decision-making component responsible for selecting and applying appropriate low-level heuristics during network operation. At each decision epoch, the controller evaluates the current network state and selects a heuristic from the pool using a heuristic selection strategy.

The selection mechanism may rely on adaptive scoring or reinforcement learning principles, where each heuristic is associated with a performance score that reflects its effectiveness under particular network conditions. After a heuristic is applied, its performance impact on network metrics is evaluated, and the corresponding score is updated.

To guide the search process, the framework incorporates an acceptance criterion that determines whether the newly generated solution should replace the current one. In this work, the acceptance decision is based on the improvement of the multi-objective performance function defined in the optimization model. If the newly produced configuration yields a lower objective function value, corresponding to improved energy efficiency, reduced delay and packet loss, and increased communication reliability, it is accepted. Otherwise, it may still be accepted with a small probability to encourage exploration and prevent premature convergence.

This adaptive mechanism allows the controller to gradually learn which heuristics perform best under different crisis network conditions.

4.3 State evaluation

Effective heuristic selection requires an accurate representation of the current network state. The state evaluation module continuously monitors key performance indicators that characterize network health and operational conditions.

The main indicators considered in this framework include:

• Residual node energy levels
• Wireless link quality and signal strength
• Traffic congestion and queue lengths
• Node density and connectivity degree
• Packet delivery performance

These indicators are aggregated into a state vector that serves as input to the hyper-heuristic controller. Based on this information, the controller identifies the most appropriate heuristic strategy to apply.

For instance, if node energy levels become critically low, the controller prioritizes energy-aware routing heuristics. In contrast, if traffic congestion increases, load-balancing heuristics may be selected to redistribute network traffic.

4.4 Acceptance and termination criteria

The hyper-heuristic search process is governed by acceptance and termination rules that ensure both exploration and convergence.

The acceptance criterion evaluates whether a newly generated network configuration should replace the current solution. Let F new denote the objective value obtained after applying a selected heuristic and F current represent the objective value of the current configuration. The new solution is accepted if

F new < F current .

In cases where the solution does not immediately improve the objective function, a probabilistic acceptance mechanism may still allow the transition, promoting exploration of alternative configurations.

The termination criterion defines when the hyper-heuristic optimization process should stop. The algorithm terminates when one of the following conditions is satisfied:

• A predefined maximum number of iterations is reached.
• The objective function shows no significant improvement over a specified number of consecutive iterations.
• Network performance reaches a stable state satisfying predefined resilience and sustainability thresholds.

These mechanisms ensure that the hyper-heuristic framework remains computationally efficient while continuously adapting network behavior to rapidly evolving crisis environments.

5 Optimization model

In crisis-resilient wireless networks, communication performance must be optimized while preserving limited energy resources and maintaining reliable connectivity. To achieve this objective, the proposed framework formulates the network management problem as a multi-objective optimization model that jointly considers energy efficiency, communication latency, packet delivery reliability, and network robustness.

Let the wireless network be represented by a graph G = ( V , E ) where V denotes the set of nodes and E represents the set of wireless communication links. The optimization objective aims to determine the most suitable communication configuration by selecting appropriate routing and communication heuristics through the hyper-heuristic framework.

5.1 Objective function

The global optimization objective is defined as a weighted multi-objective function that aggregates several key performance metrics affecting the sustainability and reliability of the network.where:

F = w 1 E + w 2 D + w 3 L + w 4 1 − R (1)

• E represents the total energy consumption of the network
• D denotes the average end-to-end packet delay
• L represents the packet loss rate
• R denotes the network reliability metric

The term ( 1 − R ) is introduced to ensure mathematical consistency with the minimization objective. Consequently, minimizing the global objective function simultaneously minimizes energy consumption, communication delay, and packet loss rate while maximizing network reliability.

The coefficients w 1 , w 2 , w 3 , w 4 are non-negative weighting factors that determine the relative importance of each objective component. These weights satisfy the normalization conditionwhich ensures a balanced contribution of each performance metric to the global objective.

∑ i = 1 4 w i = 1 (2)

The weighting coefficients were selected to provide balanced consideration between energy efficiency, communication delay, packet loss minimization, and network reliability according to the operational requirements of disaster-response communication systems. The selected weights were initially determined through empirical observations and preliminary simulation experiments in order to avoid dominance of a single optimization objective. Different weight configurations may influence routing behavior and heuristic selection decisions under varying crisis conditions. Therefore, sensitivity analysis of weighting strategies represents an important future research direction for adaptive multi-objective optimization in wireless emergency communication networks.

The proposed multi-objective formulation enables the framework to balance several conflicting communication requirements encountered in dynamic disaster environments. For example, minimizing communication delay may increase energy consumption due to additional routing activity, while improving reliability may require redundant communication paths that introduce higher network overhead. Similarly, reducing packet loss may lead to increased retransmission operations that affect both latency and energy efficiency. The adaptive hyper-heuristic framework dynamically selects suitable routing heuristics in order to maintain an effective trade-off between sustainability, communication quality, and network resilience according to real-time network conditions.

The primary goal of the optimization process is to minimize the objective function F while simultaneously maximizing communication reliability and preserving acceptable communication quality under highly dynamic crisis conditions.

5.2 Energy consumption model

The total network energy consumption is defined as the aggregate energy spent by all nodes during packet transmission, reception, and idle listening. Let V denote the set of network nodes. The total energy consumption is expressed aswhere:

E = ∑ i ∈ V E i t x + E i r x + E i idle (3)

• E i t x denotes the transmission energy consumed by node i
• E i r x denotes the reception energy consumed by node i
• E i idle represents the energy consumed during idle channel monitoring

Minimizing E directly contributes to prolonging network lifetime, which is a critical requirement in disaster environments where recharging opportunities are limited.

5.3 Communication delay

The average end-to-end packet delay measures the time required for data packets to travel from source nodes to their intended destinations. It is computed aswhere:

D = 1 P ∑ k = 1 P t k recv − t k send (4)

• P denotes the total number of successfully delivered packets
• t k send is the transmission time of packet k
• t k recv is the reception time of packet k

Reducing the delay is particularly important in emergency communication systems where timely information delivery is essential for rescue operations.

5.4 Packet loss rate

Packet loss rate reflects the reliability of the communication process and is defined as the ratio of lost packets to the total number of transmitted packets:where:

L = P lost P sent (5)

• P lost represents the number of packets that fail to reach their destination
• P sent denotes the total number of transmitted packets

Lower packet loss ensures more reliable communication between nodes during crisis response operations.

5.5 Network reliability

Network reliability reflects the ability of the communication system to maintain connectivity despite dynamic topology changes and node failures. In this work, reliability is measured through the packet delivery ratio (PDR):where P recv denotes the number of successfully received packets.

R = P recv P sent (6)

A higher reliability value indicates improved robustness of the network against link failures and congestion.

5.6 Constraints

The optimization process is subject to several operational constraints that reflect realistic network limitations.

5.6.1 Energy constraint

Each node must maintain a positive residual energy level:

E i t ≥ 0 ∀ i ∈ V (7)

5.6.2 Connectivity constraint

The network graph must remain connected to ensure communication between source and destination nodes.

G V , E is connected (8)

5.6.3 Bandwidth constraint

The data rate on each communication link must not exceed its maximum capacity:where T i j denotes the traffic load on link ( i , j ) and C i j represents the link capacity.

T i j ≤ C i j (9)

5.6.4 Cost-effectiveness consideration

In practical disaster-response deployments, communication systems must not only achieve high technical performance but also remain operationally cost-effective. Factors such as additional communication overhead, UAV deployment requirements, routing maintenance operations, and increased energy expenditure may influence the overall feasibility of large-scale emergency networking systems. Future extensions of the proposed optimization framework may therefore incorporate cost-aware objective components alongside the current sustainability and resilience metrics.

5.7 Optimization objective

The hyper-heuristic controller aims to identify the sequence of heuristic operations that minimizes the global objective function F while satisfying the above constraints. Through adaptive heuristic selection, the framework dynamically adjusts routing and communication strategies to achieve sustainable and resilient wireless networking under crisis conditions.

6 Proposed hyper-heuristic algorithm

The proposed hyper-heuristic algorithm, as shown in Algorithm 1, operates as a high-level adaptive controller responsible for selecting and applying low-level heuristics according to the observed network state. The algorithm continuously evaluates network performance indicators and dynamically adapts routing and communication strategies in order to minimize the global optimization objective defined in Section X.

The heuristic pool consists of several candidate low-level heuristics categorized into constructive, improvement, perturbation, and reconstructive strategies. These heuristics generate alternative network configurations that are evaluated according to the multi-objective function.

The hyper-heuristic controller employs a choice-function-based selection mechanism that ranks heuristics based on their historical performance and recent contribution to network improvements.

Let H = { h 1 , h 2 , … , h m } denote the set of available heuristics. For each heuristic h i, a performance score S i is maintained and updated during the search process.

At each iteration, the controller selects the heuristic with the highest choice function value:where.

C F h i = α f 1 h i + β f 2 h i + γ f 3 h i (10)

• f 1 ( h i ) represents the recent improvement produced by heuristic h i
• f 2 ( h i ) measures the historical performance of heuristic h i
• f 3 ( h i ) represents the time elapsed since the heuristic was last used
• α , β , γ are weighting parameters controlling the contribution of each component

The parameters α, β, and γ were empirically calibrated through preliminary simulation experiments in order to maintain balanced heuristic selection behavior across varying network conditions. These coefficients control the relative influence of heuristic historical performance, recent improvement contribution, and heuristic execution frequency within the choice function mechanism. To reduce stochastic effects associated with dynamic wireless environments and heuristic selection randomness, multiple simulation runs were conducted and average performance values were considered during the evaluation process.

This mechanism encourages the selection of heuristics that have previously demonstrated strong performance while maintaining exploration of less frequently applied heuristics.

After applying a selected heuristic, the resulting network configuration is evaluated using the objective function F. If the new configuration improves the objective value, it is accepted; otherwise, a probabilistic acceptance rule may still allow the transition in order to avoid local optima.

The search process continues until predefined termination conditions are satisfied.

Algorithm 1 Adaptive hyper-heuristic selection algorithm.

The termination criterion may correspond to one of the following conditions: reaching a maximum number of iterations, observing no improvement in the objective function over a predefined number of iterations, or achieving stable network performance under the specified crisis scenario.

The proposed hyper-heuristic algorithm therefore enables adaptive and intelligent selection of communication strategies, allowing the wireless network to respond effectively to rapidly evolving crisis environments while maintaining sustainable resource utilization.

7 Experimental setup

This section describes the simulation environment, the crisis scenarios used for evaluation, and the performance metrics adopted to assess the effectiveness of the proposed hyper-heuristic framework. The objective of the experimental study is to analyze the ability of the proposed approach to improve network sustainability, reliability, and communication efficiency under dynamic disaster conditions.

7.1 Simulation environment

The proposed framework is evaluated using the NS-3 network simulator, which provides a realistic packet-level simulation environment widely used for wireless networking research. The simulator enables detailed modeling of wireless communication channels, node mobility, energy consumption, and network protocols.

The simulated network consists of a heterogeneous set of wireless nodes including mobile user devices, static sensor nodes, and optional UAV relay nodes. Nodes are randomly deployed within a disaster region of size A = 1000 × 1000 m 2. The number of nodes varies between 50 and 200 in order to analyze network scalability.

Each node is equipped with a wireless interface operating under IEEE 802.11 communication standards with a transmission range of approximately 250 m. The mobility of ground nodes follows the Random Waypoint Mobility Model to emulate the unpredictable movement patterns of rescue teams and civilians in disaster areas.

The energy model is based on a battery-powered radio device where nodes consume energy during transmission, reception, and idle listening states. The initial battery capacity of each node is set to E 0 = 1000 J.

The main simulation parameters used in the experiments are summarized in Table 1.

TABLE 1. Simulation parameters.

For performance comparison, the proposed hyper-heuristic framework is evaluated against several baseline routing strategies including traditional energy-aware routing, delay-aware routing, and cluster-based communication mechanisms.

Each simulation experiment is repeated multiple times with different random seeds in order to obtain statistically reliable results.

7.2 Scenario description

To evaluate the effectiveness of the proposed framework under realistic disaster conditions, several crisis scenarios are modeled in the simulation environment.

7.2.1 Disaster-damaged urban network

In this scenario, a portion of the communication infrastructure is assumed to be destroyed due to a natural disaster such as an earthquake or explosion. Wireless nodes must establish an ad-hoc communication network to maintain connectivity among rescue teams and affected populations. Node failures and intermittent connectivity are introduced to simulate damaged communication links.

7.2.2 Emergency rescue communication system

This scenario models a large-scale emergency response operation in which mobile rescue units continuously exchange information such as victim locations, hazard alerts, and coordination messages. High traffic loads and dynamic node mobility patterns characterize this environment.

7.2.3 Dynamic connectivity disruption

In this scenario, several nodes randomly fail or become temporarily unavailable during the simulation. The network must adapt to these topology changes by dynamically reorganizing routing paths. This scenario evaluates the resilience of the proposed hyper-heuristic strategy under unstable network conditions.

These scenarios allow the evaluation of the adaptability of the hyper-heuristic framework when facing different operational constraints typically encountered in disaster environments.

7.3 Performance metrics

The performance of the proposed framework is evaluated using several widely adopted metrics in wireless networking research.

7.3.1 Network lifetime

Network lifetime measures the duration during which the network remains operational before critical nodes exhaust their energy resources. It is typically defined as the time until the first node or a significant percentage of nodes deplete their batteries.

7.3.2 Packet delivery ratio

The packet delivery ratio (PDR) measures the reliability of the communication system and is defined aswhere P sent represents the number of transmitted packets and P received denotes the number of packets successfully delivered to their destinations.

P D R = P received P sent

7.3.4 End-to-end delay

The average end-to-end delay represents the time required for a packet to travel from its source node to its destination. It includes transmission delays, propagation delays, queuing delays, and processing delays encountered along the routing path.

7.3.5 Energy efficiency

Energy efficiency evaluates the ability of the network to deliver data while minimizing energy consumption. It is measured as the ratio between successfully delivered packets and the total energy consumed by the network during the simulation.

These metrics provide a comprehensive evaluation of network performance in terms of sustainability, reliability, communication quality, and energy utilization.

8 Results and discussion

This section presents the performance evaluation of the proposed hyper-heuristic framework under the crisis scenarios described in the previous section. The objective of the experimental analysis is to assess the effectiveness of adaptive heuristic selection in improving the sustainability and reliability of wireless communication networks operating in dynamic disaster environments.

The proposed approach is compared with three conventional routing strategies: Energy-Aware Routing (EAR), Delay-Aware Routing (DAR), and Cluster-Based Routing (CBR). The evaluation focuses on key performance metrics including network lifetime, packet delivery ratio, end-to-end delay, energy consumption, and energy efficiency.

The current experimental evaluation primarily compares the proposed framework with conventional routing approaches commonly adopted in emergency wireless communication systems. Although these baseline methods provide meaningful benchmarks for evaluating communication sustainability and reliability, future studies may further extend the comparative analysis by incorporating advanced meta-heuristic optimization techniques such as Particle Swarm Optimization (PSO), Grey Wolf Optimizer (GWO), Ant Colony Optimization (ACO), and reinforcement learning-based routing frameworks. Such comparisons may provide additional insights into the effectiveness of hyper-heuristic optimization in highly complex and dynamic wireless communication environments.

To reduce stochastic bias associated with dynamic wireless network simulations and heuristic selection randomness, each simulation scenario was executed multiple times using different random seeds. The reported results correspond to the averaged performance metrics obtained across all simulation runs.

8.1 Network lifetime analysis

Network lifetime represents one of the most critical indicators of sustainability in crisis communication systems. Figure 2 illustrates the network lifetime achieved by different routing strategies as the number of nodes increases.

The results show that the proposed hyper-heuristic framework consistently achieves the longest network lifetime. This improvement is mainly due to the dynamic selection of energy-efficient communication heuristics based on real-time network conditions.

8.2 Packet delivery ratio

Packet Delivery Ratio (PDR) reflects the reliability of data transmission within the network. Figure 3 compares the packet delivery performance of the evaluated routing strategies.

The hyper-heuristic strategy achieves the highest packet delivery ratio across all network sizes. The adaptive selection of routing heuristics allows the system to avoid congested or unreliable communication paths.

8.3 End-to-end delay

End-to-end delay measures the time required for packets to travel from source nodes to destination nodes. Figure 4 illustrates the average delay observed in the simulated networks.

The results indicate that the proposed approach maintains significantly lower delays compared with conventional routing protocols. This behavior is particularly beneficial for time-sensitive emergency communication services.

8.4 Energy consumption

Figure 5 presents the total energy consumption observed during the simulation period.

The hyper-heuristic approach significantly reduces energy consumption due to its ability to dynamically switch to energy-aware routing strategies when battery levels become critical.

8.5 Energy efficiency

Energy efficiency reflects the ratio of successfully delivered packets to the total energy consumed. Figure 6 illustrates the efficiency achieved by the evaluated strategies.

The results clearly demonstrate that the proposed hyper-heuristic framework provides superior energy efficiency, making it particularly suitable for sustainable wireless communication in disaster environments.

Overall, the experimental results confirm that adaptive heuristic selection significantly improves network performance in terms of reliability, energy utilization, and communication stability. These improvements highlight the potential of hyper-heuristic optimization as an effective approach for managing crisis-resilient wireless networks operating under highly dynamic and resource-constrained conditions.

9 Conclusion

This paper presented a hyper-heuristic driven framework for sustainable and crisis-resilient wireless communication networks. The proposed approach addresses the challenges of highly dynamic disaster environments where communication infrastructure is often partially or completely unavailable. By integrating a high-level hyper-heuristic controller with a diverse pool of low-level heuristics, the framework enables adaptive selection of routing and communication strategies based on real-time network conditions.

The proposed system combines constructive, improvement, perturbation, and reconstructive heuristics to dynamically optimize network behavior. A multi-objective optimization model was formulated to simultaneously consider critical performance metrics including energy consumption, end-to-end delay, packet loss rate, and network reliability. The hyper-heuristic controller continuously monitors network indicators such as residual energy, link quality, congestion level, and node density in order to select the most suitable heuristic for each network state.

Extensive simulation experiments conducted in realistic disaster scenarios demonstrated that the proposed approach significantly improves network performance compared with conventional routing strategies. The results show that adaptive heuristic selection increases network lifetime, enhances packet delivery reliability, reduces communication delay, and improves overall energy efficiency. These improvements confirm the effectiveness of hyper-heuristic optimization for managing complex wireless networking environments under severe operational constraints.

The proposed framework therefore provides a promising solution for enabling sustainable and resilient communication systems in crisis situations such as natural disasters, emergency rescue operations, and infrastructure failures.

Future research directions include the integration of reinforcement learning mechanisms for more advanced heuristic selection strategies, the extension of the framework to UAV-assisted wireless networks, and the evaluation of the approach in large-scale real-world deployments. Further investigation may also explore distributed hyper-heuristic architectures capable of supporting ultra-dense wireless networks and next-generation emergency communication systems.

Future work may additionally investigate predictive-analysis-based routing mechanisms capable of anticipating topology changes, communication congestion, node failures, and traffic fluctuations before network degradation occurs. Integrating machine learning and predictive analytics with hyper-heuristic control may enable proactive routing adaptation and more intelligent communication management in highly dynamic disaster environments.

Moreover, future studies may incorporate noise-aware communication models and cost-effectiveness optimization criteria in order to further improve the practical feasibility and operational sustainability of crisis-resilient wireless communication systems.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: George Tsoulos, University of Peloponnese, Greece

Reviewed by: Smaranika Mohapatra, Manipal University Jaipur, India

Xiaosheng Ni, Sun Yat-sen University, China