Hybrid Secure Cluster-Based Routing Algorithm for Enhanced Security and Efficiency in Mobile Ad Hoc Networks

ABSTRACT

This research addresses critical gaps in Mobile Ad hoc Networks (MANETs) by proposing a hybrid secure cluster-based routing algorithm, focusing on enhancing network security, robustness, and reliability through multipath routing. Methodologically, the approach integrates Convolutional Neural Networks (CNN) for optimal path routing and Emperor Penguin Optimization (EPO) for clustering, introducing a novel combination for efficient cluster head selection. A novel contribution lies in the development of a prediction technique utilizing a trust assessment algorithm to calculate direct trust ratings at each node, incorporating fuzzy values between zero and one. Trust values are further influenced by node performance, adding a dynamic dimension to the trust evaluation process. Key novelties include the emphasis on energy efficiency, network longevity, remaining energy, security level, bandwidth, and packet delivery ratio as evaluation criteria. The proposed CNN-EPO model demonstrates superior results compared to traditional routing protocols, achieving a remarkable 95% energy efficiency, a heightened security level of 99%, and a throughput reaching up to 8 Mbps. Additionally, the Packet Delivery Ratio (PDR) attains close to 99% and routing overhead remains below 0.5, ensuring efficiency in challenging network scenarios with 50 adversaries. In summary, this research contributes a comprehensive solution to MANET challenges, introducing a novel hybrid routing algorithm, incorporating advanced methodologies for path optimization and clustering. These outcomes highlight how important the suggested strategy is to improve the existing state of the art in MANETs.

Introduction

High node portability and erratic wireless connectivity define Vehicular Ad Hoc Networks (VANETs), a difficult segment of Mobile Ad Hoc Networks, (MANETs) (Janani and Manikandan 2020; Xu et al. 2020). Developing a reliable navigation strategy for VANETs is crucial, given the dynamic nature of nearby vehicles and the inherent irregularities in wireless connectivity. MANETs, in general, have faced vulnerabilities due to their flexible and configurable nature, making them susceptible to attacks (BrijilalRuban and Paramasivan 2021; El-Semary and Diab 2019; Veeraiah et al. 2021). Because of its changing structure and limited funding, ad hoc networks present challenges for implementing a secure and priority Quality of Service (QoS) routing mechanism.

This study addresses these challenges through the introduction of the Movement-Aware Routing Protocol based on Hybrid Optimization (MARP-HO) for MANETs. Emphasizing the importance of secure data transfer, the research integrates advanced optimization techniques and encryption methods to enhance both energy efficiency and security in ad hoc networks (Jabbar, Saad, and Ismail 2018; Kaliyamurthi and Palanisamy 2021). The issues of energy consumption and potential cluster head failures in MANETs are tackled through a practical approach. Cooperative communication (CC) in MANETs, particularly its benefits in energy management and path selection, is explored. A hybrid inter-group operating method is proposed, leveraging cluster and place routing techniques to manage the constant movement of network nodes effectively. The study recognizes the flexibility of setting up and connecting wireless nodes in MANETs without preexisting infrastructure or centralized administration (Devi and Hegde 2018; Krishnakumar and Asokan 2022). It also highlights the limitations of traditional routing protocols that focus solely on the shortest path, overlooking factors like queuing delays and link latency. The dynamic topology of VANETs allows nodes to freely join or leave, making it a complex field of vehicle communication. The study acknowledges the added complexity introduced by transmissions and potential attacks on communication channels, including black hole or wormhole attacks. The research introduces the approach, integrating collaborative optimization with AES-128 encryption for secure routing in MANETs. For Unmanned Aerial Vehicles (UAVs) in FANETs, it prioritizes energy-efficient route design to enhance reliability and Packet Delivery Ratio. Recognizing Underwater Acoustic Networks (UWANs’) security challenges, it highlights the threat of energy leakage, emphasizing the need for stability. The study also proposes a CNN-based secure routing system for automotive ad hoc networks, crucial for spreading emergency messages in dynamic scenarios.

The proposed system innovatively combines Emperor Penguin Optimization, Convolutional Neural Networks, and robust trust management to address critical challenges in MANETs and the Internet of Things (IoT). Motivated by the need for secure, efficient, and dependable network communication, the study introduces a hybrid safe cluster-based routing algorithm. This algorithm not only ensures high throughput, security, and clustering but also dynamically selects cluster heads, progressively enhancing the network’s lifespan. The innovation lies in the integration of novel methods such as Emperor Penguin Optimization for clustering and Convolutional Neural Networks for optimal path routing. The contribution of this study is multifold, introducing a comprehensive defense mechanism against Blackhole, Wormhole, and Man in the Middle attacks. Additionally, the trust management system, utilizing TEBACA and a “rating friends” approach, ensures the selection of trustworthy cluster heads, fostering network stability. The secure routing strategy based on CNN evaluation and XOR-based encryption further fortifies the system against malicious threats. In summary, this study pioneers a holistic approach to MANET-IoT communication, offering advancements in routing, security, and trust management for a more resilient and efficient wireless network. The proposed approach holds significant industrial relevance by providing robust solutions to critical challenges in various ad hoc networks, including VANETs, UWANs, and FANETs. Its integration of advanced optimization, encryption, and neural network-based routing not only enhances network security and efficiency but also aligns with the evolving demands of industries relying on seamless and secure communication in dynamic and resource-constrained environments.

Related Work

(El-Semary and Diab 2019) proposes a clustering-based routing mechanism for MANETs, utilizing mobility information such as node degree and vehicle velocity. The circular base function neural net technique identifies cluster heads, demonstrating superior throughput, packet transmission rate, and total delay in simulations compared to other approaches (Arulselvan and Rajaram 2023) offers a hybrid trust-based secured routing system for MANETs that tackles the difficulties of environmental monitoring.The model employs the Skill Optimization Technique for node clustering, Deep Q-Learning for trust computation, and Diffusion Convolutional Neural Networks (DCNN) for routing. The approach effectively identifies and compares routing attacks, enhancing communication for environmental risk assessment (Ilakkiya and Rajaram 2024). introduces BL-IDS, a decentralized Blockchain-powered intrusion detection system for MANETs. The solution uses hybrid authentication, a similarity index for cluster formation, and wild horse optimization for routing. Results show enhanced efficiency across various metrics, demonstrating the effectiveness of the lightweight IDS in bolstering MANET performance and security.

(Karthick and Asokan 2021) presents MARP-HO, a Hybrid optimization-based mobility-aware routing algorithm for metropolitan area networks. The improved optimization for ant colonies (IACO) algorithm is used for router path calculation, while the new animal migration optimization (IAMO) technique is used for energy-efficient clustering. The evaluation shows enhanced performance across a range of metrics.

(Selvakumar and Sudhakar 2022) introduces EECSRP, an energy-efficient clustering with a secure routing protocol for MANETs. The model employs hybrid evolutionary algorithms for cluster head selection and route selection. Evaluation using NS2 shows superior performance across various parameters, affirming its efficacy in enhancing energy efficiency and reliability (Thamizhmaran and Charles 2023). focuses on CHRN-SHSP, a cluster head-based energy-efficient approach with a secure and shortest path selection routing protocol for MANETs. The model considers both obstacle and non-obstacle environments, demonstrating effectiveness in improving routing efficiency and energy management compared to existing relay node protocols (Jubair et al. 2022) proposes the QoS+ routing protocol for VANETs, addressing mobility and security challenges. QoS+ combines QoS-based CH selection and hybrid cryptography using AES and ECC. Evaluation in NS2 shows QoS+ outperforming previous works in various scenarios, with notable excellence in CH efficiency and cluster member efficiency (Rathish et al. 2021). introduces DCAIDS, a clustering-based Intrusion Detection System (IDS) for MANETs. DCAIDS employs the Weighted Clustering Algorithm (DCA) for path establishment, detecting intruder attacks, isolating affected nodes, and optimizing communication pathways.

(MVSS and Babu 2020) proposes EESSC, an energy-efficient stable and secure clustering technique for MANETs. EESSC employs fuzzy logic for CH selection and introduces Standby CH (SBCH) for improved stability and security. Simulation results demonstrate effectiveness with fewer CH changes over a 250s period (Abdulhae, Mandeep, and Islam 2022) provides a thorough analysis of cluster-based routing methods (CBRPs) for UAV-based Flight Ad-Hoc Nets (FANETs) (Deepa and Latha 2019). The review analyzes 21 CBRPs, providing insights into their strengths, weaknesses, applications, and performance measures, while identifying open issues for future research (Raj 2020) proposes a hybrid optimization model for routing in Flying Ad-Hoc Networks (FANETs), leveraging bee colony optimization (Wheeb and Naser 2021). The optimized routing strategy demonstrates superior performance by reducing delay and communication overhead compared to traditional routing protocols (Reka et al. 2023) proposes the utilization of Enhanced Chicken Swarm Optimization (ECSO) in MANETs for mobility and energy efficiency. ECSO is incorporated into the cluster head selection process of the Adaptive Position Routing Protocol (APRP), showcasing superior performance in various metrics in simulations.

(Zhou, Tan, and Iroshan 2023) proposes a novel authentication and key agreement scheme for Vehicle Ad Hoc Networks (VANETs), (Ilakkiya and Rajaram, 2023) addressing security issues. The scheme ensures mutual authentication, secure communication, and dynamic vehicle management, with efficiency evaluated to be approximately 35% lower computing and communication overhead compared to existing schemes (Zhang 2023) introduces the metaverse as a transformative technological advancement, emphasizing its potential for immersive interactions and personalized learning experiences for preschoolers (Wheeb 2022). The research highlights the significance of strong security protocols, such as encrypting and authentication using multiple factors, in guaranteeing a safe virtual space under the dynamic of the multiverse. Overall summary of existing works are shown in below Table 1.

Table 1. Summarization of related works.

Author Name	Proposed Method	Purpose	Results Metrics
El-Semary and Diab (2019)	Mobility-Based Clustering, Circular Base Function Neural Net	Develop reliable on-demand routing mechanism using AODV backbone	Throughput, Packet Transmission Rate, Total Delay
Arulselvan and Rajaram (2023)	Hybrid Trust-Based Secure Routing	Address environmental monitoring challenges	Identify and compare routing attacks (Black Hole, Gray Hole, Wormhole), Network Simulation Tools
Ilakkiya and Rajaram (2024)	BL-IDS	Decentralized Blockchain-powered IDS for MANETs	PDR, Protection Level, Identification Accuracy, Efficiency, and Time Evaluation
Karthick and Asokan (2021)	MARP-HO	Mobility-aware routing protocol for MANETs	Distribution Ratio, Loss Ratio, which Fuel Consumption, which Delay, Network Lifetime, Dead Nodes, and Efficiency
Selvakumar, M.	EECSRP	Energy-efficient clustering with secure routing for MANETs	Superior Performance in Energy Efficiency and Reliability across Various Parameters
Thamizhmaran and Charles (2023)	CHRN-SHSP	Cluster head-based energy-efficient approach for temporary networks	Improved Routing Efficiency and Energy Management in MANETs Compared to Existing Relay Node Protocols
Jubair et al. (2022)	QoS+ Routing Protocol	Address mobility and security challenges in VANETs.	Higher success rates, throughput, and efficiency.
Rathish et al. (2021)	DCAIDS	Intrusion Detection System (IDS) for MANETs.	Enhanced reliability and security; optimized pathways.
MVSS and Babu (2020)	EESSC	Energy-Efficient Stable and Secure Clustering for MANETs.	Fewer CH changes, improved stability, and efficiency.
Abdulhae, Mandeep, and Islam (2022)	CBRPs Survey	Survey on Cluster-Based Routing Protocols (CBRPs) for FANETs using UAVs.	Analysis of strengths, weaknesses, and performance measures.
Raj (2020)	Hybrid Optimization	Hybrid optimization model for routing in FANETs, leveraging bee colony optimization.	Reduced delay and communication overhead.
Reka et al. (2023)	ECSO in MANETs	Enhanced Chicken Swarm Optimization (ECSO) in MANETs for mobility and energy efficiency.	Superior performance in energy efficiency and packet rates.
Zhou, Tan, and Iroshan (2023)	VANETs Security Scheme	Authentication and key agreement scheme for VANETs addressing security vulnerabilities.	Lower computing and communication overhead; practicality.
Zhang (2023)	Metaverse Exploration	Exploration of metaverse for interactive learning experiences and addressing cybersecurity concerns.	Emphasis on encryption, multi-factor authentication, audits.

Proposed Work

System Model

Naeem et al., (2023) The proposed system model has n mobile nodes that are free to roam about the network, such as N_1,N_2,…,N_n. The minimum and maximum mobility of the nodes is denoted by [minμ,maxμ], and the nodes have various velocities. On the basis of distance and movement metrics, the whole network is divided into several clusters. The nodes that are closest to one another and have comparable mobility are grouped together.

In Figure 1, the suggested MANET-IoT is seen. The internet includes both certified and untrusted nodes, which are as depicted in the picture (malicious or attacker) (Wheeb et al., 2023). Presumably, there are always fewer assailant nodes (m) than normal nodes (n), or m<n. Table 1. Provides a summary of the danger models under consideration.

Figure 1. Proposed network model.

Each MANET is a technology needs effective forwarding to handle the potential for network failures, attempts at hacking, and constraints on resources. This study proposes a method for a hybrid safe cluster-based routing algorithm, delivering high throughput, security, clustering, and picking cluster heads, in order to make it possible to gradually increase the network’s life duration. In this analysis, the results are contrasted with the conventional routing protocol. At first, CNN is considered for optimal route routing, while EPO is demonstrated for clustering. Both of these are offered as first methods. Our method of prediction utilizes a trust evaluation algorithm at each node, which is then normalized as a fuzzy number between zero and one. This allows us to determine the direct trust rating. In addition to this, the trust value is determined by the performance of each node. The experimental results are evaluated based on a number of parameters, include energy efficiency, network durability, residual energy, level of security, bandwidth, and the percentage of packets delivered. According to the investigation, the suggested routing algorithm has the ability to offer MANET-IoT a lot more organized, longer network lifetime, higher quality of service, and more cost-effective routes than the current approaches.

Threat Model

This work detects and mitigates the following attacks:

• Blackhole: The attacker injects routing sensor data into the network, sends packets in the direction of the problem, and then drops them. The attacker in a black hole attack will brag to the defenders that their route is the quickest. If this answer comes in before the true one, a false route including the harmful network node will be constructed. A packet may now be dropped by this malicious node. The throughput and packet-to-delivery ratio of the AODV protocol may be evaluated using a black hole attack by stationing an attacker on a single node.

• Wormhole: In a worm hole attack, a malicious node seizes data from one section of the network and sends it via a hidden tunnel to another. Disruptions in the network may occur if information regarding fault routing leaked. Using encryption and node location data, the authors of established a way to defend MANET from this assault.

Attack known as the “Man in the Middle”

This type of assault involves a malevolent node positioned in between the source and the target. After that, it gathers all of the messages and either alters or deletes them. Since MANET transfers data hop-by-hop, it is susceptible. Both authentication and encryption provide robust defenses against such an attempt.

Clustering

The cluster’s leader amasses all the information and authority from the group’s members. This data is utilized to identify the most appropriate cluster head for transmission to the command center. The information collected from mobile nodes will be used by the ground station to decide which nodes will act as CHs. Here, the cluster head is identified in terms of the network’s health using Emperor penguin optimization. The best nodes may be clustered using this method. The basics of optimizing for the Emperor Penguin are as follows:

• Set the initial huddle perimeter.

• Determine the temperature distribution in the area of the group

• Determine the fitness value.

• The ideal fitness function may be attained by relocating the best mover.

Set the Huddle Boundary to Zero.

The imperial penguin will determine the first huddle boundary according to the minimum & maximal range of a wireless sensor network. Boundary values vary with context. The perimeter of the cluster is in the shape of an L. The emperor penguin serves as the focal point of the L-shaped polygon. As a Emperor Penguin makes a move, the others all shift to follow suit. The most efficient worker is selected in this group discussion. We pinpoint the most influential driver in forming clusters of functional sensors. A number of emperor penguins are dispersed over the shape. The emperor penguin shares its habitat with at least 2 other species of penguin. Despite its rapid appearance, the penguin is really much slower than the wind.

The wind speed is used to determine the Huddle limit for the emperor penguin $V_w$ and its gradient${\phi }_w$. Here, we provide the correlation between wind speed and slope in Equation 1(1) ${\phi }_w=\mathrm{\nabla }{V}_w$(1) .

(1) ${\phi }_w=\mathrm{\nabla }{V}_w$(1)

Hence, the initialization of the huddle border is as Equation 2(2) $F_w={\phi }_w+\mathit{i}\mathrm{\Omega }$(2) , where v is the wind velocity.

(2) $F_w={\phi }_w+\mathit{i}\mathrm{\Omega }$(2)

Calculating the Temperature Profile

The temperature is a major consideration when it comes to exploring & profiting just on huddle’s search space. The penguins are huddling together to stay warm, with the dominating emperor penguin in the center. The vibe of the group embrace $T_H$ is depends on the radius of the huddle $R_H$. It is given by the following Equations 3(3) $T_H=0;\mathit{if}{R}_H>1$(3) and 4(4) $T_H=1;\mathit{if}{R}_H<1$(4) .

(3) $T_H=0;\mathit{if}{R}_H>1$(3)

(4) $T_H=1;\mathit{if}{R}_H<1$(4)

The temperature profile is adjusted after each repetition by solving the following Equation 5.(5) $T_H^{\prime }=\mathit{T}-\frac{I_m}{\mathit{i}-I_m}$(5)

(5) $T_H^{\prime }=\mathit{T}-\frac{I_m}{\mathit{i}-I_m}$(5)

The iteration number I is represented in the preceding equation by the term I and $I_m$ as many times as possible. With the help of the three equations, the temperature profile at the huddle border is adjusted.

Analysis of Fitness Function Criterions

Analysis of Energy Consumption

The total energy $E_t$ consumed as Equation 6(6) $E_t\left(s_i\mathit{\hspace{0.17em}},\mathit{d}\right)=\left\{\begin{array}{c}{s}_i\mathit{E}+s_i{\epsilon }_{\mathit{fs}}{d}^2,\mathit{d}\le d_0\\ s_i\mathit{E}+s_i{\epsilon }_{\mathit{mp}}{d}^2,\mathit{d}>d_0\end{array}\right.$(6)

(6) $E_t\left(s_i\mathit{\hspace{0.17em}},\mathit{d}\right)=\left\{\begin{array}{c}{s}_i\mathit{E}+s_i{\epsilon }_{\mathit{fs}}{d}^2,\mathit{d}\le d_0\\ s_i\mathit{E}+s_i{\epsilon }_{\mathit{mp}}{d}^2,\mathit{d}>d_0\end{array}\right.$(6)

Where $d_o=\sqrt{\frac{{\mathit{\epsilon }}_{\mathit{fs}}}{{\epsilon }_{\mathit{mp}}}}$

Where E indicates the energy used to forward or accept one bit message; ${\epsilon }_{\mathit{fs}}$ is the amplification coefficient of free-space signal and ${\epsilon }_{\mathit{mp}}$ is the multi-path fading signal intensification coefficient, its rates based on the path amplifier model and d indicates the distance among transmitter and receiver; $s_i$ is the bit quantity of transmitting data packets.

Mobility Prediction

Velocity or stability is a vital factor in decide the CHs. To avoid frequent CH change this present scheme is wanted to select a CH that does not travel very rapidly. Whether the CH node travels quick means the nodes might be dividing from the CH and as an output of a re-affiliation occur. This can enlarge computation and processing, that does not an attractive characteristic. The stability of the connection will also increase the strength of the group. The node’s displacement along the rate of that displacement indicate the stability of every connation. Equation 7(7) $\mathit{LD}=\frac{-\left(\mathit{ab}+\mathit{cd}\right)+\sqrt{{\left(a^2+c^2\right)}^2-{\left(\mathit{ad}-\mathit{bc}\right)}^2}}{a^2+c^2}$(7) can be used for estimating the duration of any link.

(7) $\mathit{LD}=\frac{-\left(\mathit{ab}+\mathit{cd}\right)+\sqrt{{\left(a^2+c^2\right)}^2-{\left(\mathit{ad}-\mathit{bc}\right)}^2}}{a^2+c^2}$(7)

Where, $\mathit{a}=v_i\mathit{cos}{\theta }_i-v_j\mathit{cos}{\theta }_j,\mathit{b}=x_i-x_j,\mathit{c}=v_i\mathit{sin}{\theta }_i-v_j\mathit{sin}{\theta }_j,\mathit{b}=y_i-y_j$ and $v_i,$ $v_j$are speeds of mobility. Some GPS data is required to determine the node’s level of stability. After estimation, LD (Link Delay) is reverse. stability of a node-to-node communication as measured by Equation 8.(8) $\mathit{L}{N}_s=\mathit{e}-\mathit{LD}$(8)

(8) $\mathit{L}{N}_s=\mathit{e}-\mathit{LD}$(8)

Where, the Minimum rate of $N_s$ offers extra stable connection.

Evaluate the Fitness Function

The amount of power that a sensor node consumes is the fitness function in this investigation. In an attempt to reach its full fitness potential, an imperial penguin is marching toward its fellow species’ representative in the center. The penguin constantly changes its posture based on the most efficient movers it has identified. The current emperor penguin population size is compared to the original emperor penguin population size to determine the position update.

The distances between emperor penguins are determined by the following factors:

• Things that make the greatest emperor penguin inevitable $S_{\mathit{Ep}}$.

• Considerations for avoiding collisions $A_{\mathit{EP}}$ and $C_{\mathit{EP}}$.

• The most effective penguin posture $P_B$

• Where are emperor penguins right now $P_{\mathit{Ep}}$

• Present version i.

Using these factors, we can solve for the distances between penguin using Equation 9.(9) $\mathit{D}=\mathit{abs}\left(S_{\mathit{Ep}}\left(A_{\mathit{EP}}\right).P_B\left(\mathit{i}\right)-C_{\mathit{EP}}.P_{\mathit{Ep}}\left(\mathit{i}\right)\right.$(9)

(9) $\mathit{D}=\mathit{abs}\left(S_{\mathit{Ep}}\left(A_{\mathit{EP}}\right).P_B\left(\mathit{i}\right)-C_{\mathit{EP}}.P_{\mathit{Ep}}\left(\mathit{i}\right)\right.$(9)

Both of these elements are critical to the emperor penguin’s social structure $\mathit{f}$ and 1. The emperor penguin’s forays into and use of the huddle area are driven by these two criteria. Equation 10(10) $S_{\mathit{Ep}}=\left(\mathit{f}.e^{-\mathit{i}}/\mathit{l}-e^{-\mathit{i}}\right)$(10) below provides the formula.

(10) $S_{\mathit{Ep}}=\left(\mathit{f}.e^{-\mathit{i}}/\mathit{l}-e^{-\mathit{i}}\right)$(10)

The crash evading features $A_{\mathit{EP}}$ and $C_{\mathit{EP}}$ is considered spending the following Equations 11(11) $A_{\mathit{EP}}=\left(2\ast \left(T_H^{\prime }+P_{\mathrm{grid}}\left(\mathrm{accuracy}\right)\ast \mathrm{rand}\right)\right)-T_H^{\prime }$(11) –13(13) $C_{\mathit{EP}}=\mathrm{rand}\left(0,1\right)$(13) .

(11) $A_{\mathit{EP}}=\left(2\ast \left(T_H^{\prime }+P_{\mathrm{grid}}\left(\mathrm{accuracy}\right)\ast \mathrm{rand}\right)\right)-T_H^{\prime }$(11)

The $P_{\mathrm{grid}}$ The optimum emperor penguin location minus the actual emperor penguin location is the measure of accuracy.

(12) $P_{\mathrm{grid}}\left(\mathrm{accuracy}\right)=P_B-P_{\mathit{Ep}}$(12)

A random number between 0 and 1 is used as the collision avoidance factor.

(13) $C_{\mathit{EP}}=\mathrm{rand}\left(0,1\right)$(13)

To calculate the difference between the current empire penguins and the best imperial penguin, substitute Equation (Eq 10(10) $S_{\mathit{Ep}}=\left(\mathit{f}.e^{-\mathit{i}}/\mathit{l}-e^{-\mathit{i}}\right)$(10) , 11(11) $A_{\mathit{EP}}=\left(2\ast \left(T_H^{\prime }+P_{\mathrm{grid}}\left(\mathrm{accuracy}\right)\ast \mathrm{rand}\right)\right)-T_H^{\prime }$(11) , 12(12) $P_{\mathrm{grid}}\left(\mathrm{accuracy}\right)=P_B-P_{\mathit{Ep}}$(12) , & 13(13) $C_{\mathit{EP}}=\mathrm{rand}\left(0,1\right)$(13) in Eq 6(6) $E_t\left(s_i\mathit{\hspace{0.17em}},\mathit{d}\right)=\left\{\begin{array}{c}{s}_i\mathit{E}+s_i{\epsilon }_{\mathit{fs}}{d}^2,\mathit{d}\le d_0\\ s_i\mathit{E}+s_i{\epsilon }_{\mathit{mp}}{d}^2,\mathit{d}>d_0\end{array}\right.$(6) . Hence, the effective movers are identified, and the imperial penguins’ distributions are revised.

Relocation of the Emperor Penguin

At each iteration, the emperor penguins move about until they find the best spot. The most recent information on the emperor penguin’s location is as follows Eq. 14(14) $P_{\mathit{Ep}}\left(\mathit{i}+1\right)=P_{\mathit{Ep}}\left(\mathit{i}\right)-A_{\mathit{EP}}.\mathit{D}$(14) .

(14) $P_{\mathit{Ep}}\left(\mathit{i}+1\right)=P_{\mathit{Ep}}\left(\mathit{i}\right)-A_{\mathit{EP}}.\mathit{D}$(14)

The emperor penguin’s position will be constantly updated when the loop is severed. The optimal position of a emperor penguin reveals which cluster members are active, and its value reveals the vitality of the nodes.

After the stopping requirement is reached, an emperor penguin will stop executing Eqs 3(3) $T_H=0;\mathit{if}{R}_H>1$(3) -11(11) $A_{\mathit{EP}}=\left(2\ast \left(T_H^{\prime }+P_{\mathrm{grid}}\left(\mathrm{accuracy}\right)\ast \mathrm{rand}\right)\right)-T_H^{\prime }$(11) . Using Emperor penguin optimization, the best cluster heads for a MANET-IoT may be identified. On the basis of these cluster masters, the reliable CN is employed for data transport.

Table

Algorithm 1 for Cluster Head Selection

Set up the transmission parameters for the wireless sensor network. set the optimisation parameters for the Emperor Penguin, including $V_w,T_H,S_{\mathit{Ep}}$, $A_{\mathit{Ep}},C_{\mathit{Ep}},P_{\mathit{Ep}}$ and Iteration I. network energy and data to determine the fitness function value. Use Eq 2(2) $F_w={\phi }_w+\mathit{i}\mathrm{\Omega }$(2) as a starting point for the huddle border. Use Equation 5(5) $T_H^{\prime }=\mathit{T}-\frac{I_m}{\mathit{i}-I_m}$(5) to get the temperature at the huddle’s perimeter. Using Eqs 6(6) $E_t\left(s_i\mathit{\hspace{0.17em}},\mathit{d}\right)=\left\{\begin{array}{c}{s}_i\mathit{E}+s_i{\epsilon }_{\mathit{fs}}{d}^2,\mathit{d}\le d_0\\ s_i\mathit{E}+s_i{\epsilon }_{\mathit{mp}}{d}^2,\mathit{d}>d_0\end{array}\right.$(6) -10(10) $S_{\mathit{Ep}}=\left(\mathit{f}.e^{-\mathit{i}}/\mathit{l}-e^{-\mathit{i}}\right)$(10) , determine how far apart the emperor penguins are. Use equation (Eq.) 11 to update the emperor penguins’ location. Find the optimal answer, and make sure the result is current. Determine the value of the fitness function at the new location, and then update the results. The limit of allowed iterations must be checked. If it is, go to the next step. Otherwise, it must be repeated stage 5. The optimal answer gives back the power from the node with the most weight. The coordinates indicate the preferred cluster head.

Trust Management

The routing procedure starts off with the determination of each cluster’s head. A trustworthy hub is required of the cluster’s leader. No malicious, self-serving, or gray hole-attacking nodes may be elected as cluster leaders. Hence, controlling trust is essential. The most reliable nodes in the cluster may be identified with the use of a challenge evaluation process i.e $T_{\mathit{CHALLENGE}}$, In order to strengthen the network’s stability and collect information about each other for a friends list, one possible technique is a “rating friends” one, in which each node ranks its nearby neighbors. During the friend-sharing phase, node A will ask node B for its buddy list and then update its own list accordingly. If nodes A or node B are friends, and if node C is also a neighbor of node B yet not of nodes A, then node A may determine its confidence in node C by node B. B i.e $T_{\mathit{RECOMMENDED}}$.

Each mobile site in a cluster is given a trustworthiness score based on the parameters listed above. The honesty, proximity, collaboration, and energy reserves of a node are the determining factors in its reliability (i.e., values of QoS trust metrics). Hence, TEBACA is relied upon, since it combines information from its own ideas and its interactions with its neighbors into a single score is shown in Eq 15(15) $\begin{array}{cc}\mathit{\hspace{0.17em}}& \mathrm{Trust},\mathit{T}=\left(w_1\times E_{\mathrm{REMAINING}}\right)\\ \mathit{\hspace{0.17em}}& +\left(w_2\times \mathrm{ceil}\left(T_{\mathrm{CHALLENGE}}\times 10\right)\right)\\ \mathit{\hspace{0.17em}}& +\left(w_3\times \mathrm{ceil}\left(T_{\mathit{RECOMMENDED}}\right)\right)\end{array}$(15) .

(15) $\begin{array}{cc}\mathit{\hspace{0.17em}}& \mathrm{Trust},\mathit{T}=\left(w_1\times E_{\mathrm{REMAINING}}\right)\\ \mathit{\hspace{0.17em}}& +\left(w_2\times \mathrm{ceil}\left(T_{\mathrm{CHALLENGE}}\times 10\right)\right)\\ \mathit{\hspace{0.17em}}& +\left(w_3\times \mathrm{ceil}\left(T_{\mathit{RECOMMENDED}}\right)\right)\end{array}$(15)

A node on a friend list may have a trustworthiness between 0 and 10. If a node’s value is higher above 10, it is considered hostile. Hence, bonds formed between friends are held in high regard. Nodes that perform very well in three metrics – residual Energy, network throughput, and node connection – are selected as cluster leaders. Due to the dynamic nature of a cluster’s nodes, we offer a method for periodically updating the cluster’s trustworthiness. With this updated trust value, we’ll be more able to identify attackers and select a new cluster leader.

When every one of the nodes has completed their calculations, they will then share their neighborhood views with their neighbors, or the gadgets within a hop count of each other. The local perspectives of a node’s neighbors may influence the node’s own viewpoint. This method takes a simplistic tack by averaging out all of the regional viewpoints. Within the equation $\mathit{n}1$ represents node and $T_1$ represents old trust value of node and $T_2$ represents changed value.

Table

Algorithm 2 for Trust Update

Start Inputs of $n_1,T_1$ Outputs A, GV For for each node $n_i$ Program $T_1$ to all its inhabitants Upon reaction of ${\mathrm{T}}_1$ from node $n_1$: GOTO subroutine: Inputs of $n_1,T_1$ Outputs of $n_1,T_1{ }^{\mathrm{\prime }}$ Upon response of ${\mathrm{T}}_2$ from node $n_1$ If $T_2\ne T_1$ Merge the $T_2$ and $T_1$ based on the following conditions: {node $n_i$ is in both $T_2$ and $T_1$, find the mean of the columns that match to $n_i$ in both $T_2$ and $\left.T_1.\right\}$ Store the average value of node $n_i$ in an in-between list $\mathrm{TEM}{\mathrm{P}}_i$. If {node $n_i$ is in either $T_2$ or $\left.T_1\right\}$ End

Secure Routing

The increasing popularity of mobile and IoT applications has increased the need for a reliable wireless network routing mechanism [25]. This is because traditional routing methods can’t improve routing depending on how well a network has performed in the past. To this end, we suggest a CNN-based routing strategy is shown in Figure 2. We introduce a routing approach that decides on routes depending on how well the CNN thinks the network is doing. The characteristics of the flow and security systems, in addition to the intermediate nodes, are added to the network feature matrix, which improves the CNN’s evaluation of the network’s performance.

Figure 2. CNN-Based routing strategy for enhanced network performance evaluation.

The variables are given as Eq 16(16) $\mathit{SK}=\left(\mathit{S}{K}_1,\mathit{S}{K}_2,\mathit{S}{K}_3,\mathit{S}{K}_4,\mathit{S}{K}_5,\mathit{S}{K}_6,\mathit{S}{K}_7,\mathit{S}{K}_8\right)$(16) –18(18) $\mathit{IV}=\left(\mathit{I}{V}_1,\mathit{I}{V}_2,\mathit{I}{V}_3,\mathit{I}{V}_4\right)$(18) ,

(16) $\mathit{SK}=\left(\mathit{S}{K}_1,\mathit{S}{K}_2,\mathit{S}{K}_3,\mathit{S}{K}_4,\mathit{S}{K}_5,\mathit{S}{K}_6,\mathit{S}{K}_7,\mathit{S}{K}_8\right)$(16)

(17) $\mathit{R}=\left(R_1,R_2,R_3,R_4,R_5,R_6,R_7,R_8\right)$(17)

(18) $\mathit{IV}=\left(\mathit{I}{V}_1,\mathit{I}{V}_2,\mathit{I}{V}_3,\mathit{I}{V}_4\right)$(18)

It works upon operations of 16-bit words. To safeguard the data, restricted OR actions and addition modulus procedures are used [26]. The plaintext portion (PT_i), or data packet, is secured in the following way: Equation 19.(19) $\mathit{C}{T}_i=\mathit{WD}16\left(R_4^i⊞t_3,K_5,K_6,K_7,K_8\right)⊞R_1^i$(19)

(19) $\mathit{C}{T}_i=\mathit{WD}16\left(R_4^i⊞t_3,K_5,K_6,K_7,K_8\right)⊞R_1^i$(19)

Then, this ciphertext ($\mathit{C}{T}_i$) is transmitted through optimal route.

Table

Algorithm 3 for Secure Routing

(1) Start (2) Initialize available routes (3) For each ($R_i$) (4) Compute all three metrics (5) Select optimal route ($R_{\mathit{Best}}$) (6) End For (7) Initialize $\mathit{SK},\mathit{IV},\mathit{t}$ using Eq 16(16) $\mathit{SK}=\left(\mathit{S}{K}_1,\mathit{S}{K}_2,\mathit{S}{K}_3,\mathit{S}{K}_4,\mathit{S}{K}_5,\mathit{S}{K}_6,\mathit{S}{K}_7,\mathit{S}{K}_8\right)$(16) –18(18) $\mathit{IV}=\left(\mathit{I}{V}_1,\mathit{I}{V}_2,\mathit{I}{V}_3,\mathit{I}{V}_4\right)$(18) (8) Encrypt data ($\mathit{C}{T}_i$) using Eq. 19(19) $\mathit{C}{T}_i=\mathit{WD}16\left(R_4^i⊞t_3,K_5,K_6,K_7,K_8\right)⊞R_1^i$(19) . (9) Transmit $\mathit{C}{T}_i\to \mathit{Destination}$ (10) End

Results and Discussion

Simulation Setup

The simulation program Network Simulator Version 3 (NS-3.25) is used to construct the suggested MANET network [28]. The events network management simulator, ns-3.25, is used for setting up the network using the computer dialects C++ and TCL. As a matter of fact, ns-3.25 allows for the integration of Blockchain and implements most wireless protocols used for connectivity. As a result, we used ns-3 for computations. Table 2 displays the key approximated variables.

Table 2. Simulation parameters.

Factor		Rate
Network Area		1000*1000 m
Count of Clusters		10 (max)
Count of nodes		100
$\mathit{PuK}$ size		32 KB
Initial energy of nodes		1000 J
Network bandwidth		20 MHz
Count of packets		1000
Size of the packet		32 KB
Count of retransmissions		7 (max)
Mobility range	${\mu }_L$	20 m/s
	${\mu }_U$	40 m/s
Maximum Trust value		100
$\mathit{\alpha }$		0.1
$\mathit{\beta }$		0.3
$\mathit{\gamma }$		0.5
Number of attackers		25
Simulation time		100 s

By a network with cloud server, a hundred nodes are taken into account, as shown in Table 2.

Comparative Analysis

The effectiveness of the suggested system is assessed using a variety of measures, such as the use of power, power remained the ratio of packet delivery (PDR), productivity, network lifespan, routing above, and security. We analyze historical efforts in comparison to where AODV, Trust Routing, and E-MAT are right now [27].

Analysis on Energy Efficiency

The efficiency with which network nodes utilize the energy given to them is quantified by the fraction of that energy that is actually used.

Figure 3. compares the thermal efficiency of the projected and current works. The desired effort is preferable to the alternates, as the outcomes demonstrate. The proposed model employs a method of route selection that maximizes the energy efficiency of the network and can bring that efficiency up to 95%. This is far better than existing AODV routing and trust based routing. In AODV, the route is selected only based on the availability while trust-routing selects route based on the trust value alone. Due to this limited metric consideration the existing works fail to achieve energy efficiency below 80%. Thus, involvement of optimum cluster formation and route selection improves the overall energy efficiency. The main reason behind this result is involvement of cluster formation and optimal routing in the proposed approach [29].

Figure 3. Comparison on energy efficiency.

Analysis on Network Lifetime

When a network has living links with leftover energy above its foundation, it is said to have a “lifetime.”The number of nodes alive for the specified time period is compared in Figure 4. The total number of live nodes steadily decreases as the time in seconds increases. There are 100 nodes alive when the time is first started, or 1 second. As the period is extended, this number decreases. In the course of trust-based forwarding strategy, AODV has just twenty active nodes while the scheme with 30 live nodes is used. Nevertheless, in 100s of hours of simulation, 70 nodes are still alive in the suggested task.

Figure 4. Comparison on number of nodes alive.

Figure 5(a,b) compares the total amount of nodes and the total number of attackers that have attacked the network over its existence. The lifespan of the system rises in accordance to the link count. On the other hand, a higher number of attackers results in a shorter network lifetime. In other words, an attacker’s engagement in the network significantly reduces its longevity. When 10 attacker nodes are involved, the suggested network can withstand 75 rounds, compared to 9 rounds with the AODV and 18 rounds for Trust-based route. Thus, the proposed methodology is far better than existing works in providing better network lifetime by controlling attacks involvement in the network [30].

Figure 5. Comparison on network lifetime.

Analysis on Residual Energy

Each network node’s current energy condition is described by the critical performance metric known as residual energy.

Figure 6, 7, compares the projected and existing works’ energy efficiency. Energy utilization often increases when there are more attackers on the network. The energy supplies of the nodes are being attacked. The nodes will ultimately stop functioning after a given amount of time when their energy is completely used up. To prevent this, the entire network has to be protected against attacks.

Figure 6. Comparison on residual energy.

Figure 7. Comparison on security level.

Analysis on security level

The degree to which packets in a network have been modified is indicative of its susceptibility to attack. The simple fact that numerous information packages are changed suggests that there are multiple offenders on the network at once.

The research indicates that the work suggested has a degree of safety of 99%, which is far higher than the ones that that are already in use, as figure 3.7 illustrates. The recommended work’s security is increased by the following features: (i) a network’s multi-attribute cryptography provides excellent safety and effectively uses identification to prevent unwanted nodes from joining. Consequently, there are no unsafe or unapproved nodes in the network. (ii) To ensure that there are no dangerous untrustworthy nodes, the multifaceted trust value is taken into account when choosing the best course. Encrypted secures data by preventing rogue nodes from listening in on it or altering it.

Analysis on Throughput

Throughput measures how much information can be transmitted from one location to another during a certain period of time through a network. Bits, rather than megabits per second or gigabits per second, are often used to quantify network performance (bps).

Figure 8 displays a contrast between the projected and actual outputs of the works. Throughput, influenced by channel security and data transmission methods, exhibits a twofold improvement over existing works, reaching up to 8 Mbps. This success is attributable to the recommended works strong managing networks using routing methods and safety plan. The approach not only outperforms in data transmission but also significantly minimizes data loss. Further dissecting throughput with respect to nodes (Figure 8a) and attackers (Figure 8b), the figures indicate a progressive decline in throughput as the number of attacker’s increases, aligning with expectations due to increased attacks on data transfers. Despite this, the suggested approach consistently outshines current works, maintaining optimal throughput even under challenging scenarios.

Figure 8. Comparison on throughput.

Analysis on Packet Delivery Ratio (PDR)

The percentage of sent communications that actually arrive at the address they were intended is known as the packet arrival ratio, or PDR.

As Figure 9 illustrates, the PDR produced by the planned work is very nearly 99%, which is much greater than the PDR produced by the ongoing programs. In case 1, the PDR increases steadily as the node count increases. The probability that the best path will be chosen rises with the amount of nodes. Consequently, the data is transmitted efficiently and consistently. Conversely, the PDR decreases as the number of assailants rises. This is due to increased attack activity that is modifying route data and leading to packet loss. Nonetheless, the suggested Block-Sec approach yields superior PDR in both cases. This study demonstrates that, in spite of the recommended security precautions, cluster-based routing leads to optimum routing.

Figure 9. Comparison on PDR.

Analysis on Routing Overhead

Routing overhead is caused by the extra management packets required to transmit information packets effectively. It is the ratio of the overall quantity of traffic control messages sent from all of the nodes to the amount of data packets received at the destination node.

Figure 10 depicts the association between the frequency of network nodes and the number of attackers and the route overhead of both current and suggested solutions. It has been demonstrated that in a distributed system with a unique identifier and 50 competitors, the recommended technique results in less than 0.5 of a networking overhead. Retransmissions that route demand generation become necessary when the number of attackers on the E-MANET increases, adding to the complexity. Conversely, the suggested approach preserves the best broadcast path based on a multi-modal route, which reduces the amount of repeat transmissions, or routing overhead.

Figure 10. Comparison on routing overhead.

Discussion

The experimental results provide valuable insights into the proposed strategy’s performance. With a meticulous focus on objectivity and conciseness, we aim to enhance the discussion of these outcomes. The innovation lies in a hybrid cluster-based routing algorithm, leveraging Convolutional Neural Networks (CNN) for optimal path determination and Emperor Penguin Optimization (EPO) for efficient clustering. It’s crucial to emphasize the unique amalgamation of CNN and EPO, enabling superior throughput, security, and energy efficiency. The strategic route selection mechanism, maximizing network energy efficiency up to 95%, stands out against traditional AODV and trust-based routing approaches, which fall short below 80%. Further detailing the algorithm’s intricacies and highlighting the distinct roles of CNN and EPO in cluster formation and route optimization incorporated for a more comprehensive understanding. Furthermore, it is critical to analyze the impact of IoT principles on our suggested system. Our approach demonstrates substantial improvements in wireless network performance by incorporating IoT principles such as data-driven decision-making and adaptive networking. IoT-enabled devices add to the network’s dynamic character by providing real-time data collecting and analysis for better routing decisions and resource allocation. Our simulation includes IoT-related characteristics such as data volume, data rate, and device mobility to realistically reflect real-world IoT installations. This assures that our findings apply to IoT-enabled wireless networks, where scalability, dependability, and energy efficiency are critical. In conclusion, our proposed system offers a big step forward in wireless network optimization, utilizing novel methodologies and IoT principles to improve performance, dependability, and security. By solving fundamental MANET difficulties and embracing IoT concepts, our method paves the path for more efficient and resilient wireless networks in IoT settings.

Conclusion & Future Work

This study proposes a method for a hybrid safe cluster-based routing protocol, delivering high bandwidth, security, clusters, and picking cluster heads, in order to make it possible to gradually increase the network’s life duration. In this analysis, the results are contrasted with the conventional routing protocol. At the beginning, convolutional neural networks are introduced for the purpose of optimal route routing, while EPO is introduced for the purpose of clustering. Our method of prediction utilizes a trust evaluation algorithm at each node, which is then normalized as a fuzzy number between zero and one. This allows us to determine the direct trust rating. In addition to this, the trust value is determined by the performance of each node. The experimental results are evaluated based on a number of parameters, achieving a 99% security level, 8 Mbps throughput, and 99% Packet Delivery Ratio, the impact of IoT is evident in extended network lifetime and enhanced energy efficiency (up to 95%). The research indicates that the proposed routing algorithm has the potential to provide sensors in wireless networks a greater service quality, a longer network lifetime, more energy-efficient routing, as well as more secure routing than the protocols that are now in use. Future study aims to use blockchain for enhanced security in MANET-based IoT, taking advantage of its decentralized nature. Furthermore, unique optimization strategies for IoT-enabled MANETs will improve network performance. These efforts aim to improve wireless network optimization and security while addressing unique problems and promoting scalability.

Ethics Approval and Consent to Participate

No participation of humans takes place in this implementation process.

Human and Animal Rights

No violation of Human and Animal Rights is involved.

Authorship Contributions

All authors are contributed equally to this work.

Acknowledgements

There is no acknowledgement involved in this work.

Data availability statement

Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.