Table 1 A comparison of existing clustering and routing approaches in MANETs.

Jose et al.¹

Proposed F-CAPSO using fuzzy-chaos PSO and SSARM-SCA for clustering and routing.

Enhances energy efficiency and improves secure data transmission.

Does not address adversarial attacks or blockchain-related metrics.

Shahbazi et al.²

AI-driven CH selection using modified DBSCAN with weighted flight metrics

Improves PDR, reduces latency, and extends CH lifespan

Performance depends on accurate and frequent aircraft data updates

Hussein et al.³

Proposed a Coverage K-Means Routing Protocol (CKRP) using zone-based clustering and optimized CH election.

Enhances route reliability and minimizes latency in dense MANETs.

No explicit results on energy consumption or scalability under attack scenarios.

Ahirwar et al.⁴

Proposed SE2CG-ORP using FS_TES encryption, K-Means clustering, T2_FC for CH selection, and CGO for path optimization in MANETs

Enhances security and energy efficiency; superior PDR, residual energy, and throughput; optimized latency and encryption/decryption times.

Lack of detailed evaluation under attack scenarios and limited validation metrics on adversarial robustness and memory overhead.

Saminathan et al.⁵

Introduced MRCP-DNID-MANET-DONN using DONN, SCGC, and ARO for IDS in MANETs

Enhances intrusion detection with high accuracy, lower energy usage, and multipath routing.

Specific values like intra-cluster distances or latency per hop not provided.

Reka et al.⁶

Proposed COCG-MSA-GCNN-MA-ID-MANET combining Coati Optimization for clustering and MSA-GCNN for IDS.

Achieves high detection accuracy, low memory use, and low computation time.

High computational complexity of deep learning and lack of labeled multi-attack MANET datasets.

Ravindran et al.⁷

SECOA integrates RNN-based link lifetime prediction with Coati Optimization

Achieves high trust, low latency, and optimal energy utilization in routing

Limited discussion on scalability under large-scale or adversarial conditions

L.V. et al.⁸

Introduced CCCH with centralized Cluster Coordinator managing CH selection

Minimizes CH changes and energy consumption while improving PDR

Does not evaluate model under large-scale or attack-prone environments

Nirmaladevi et al.⁹

Hierarchical clustering with fuzzy crow search-based CH selection, authentication-based selfish node detection, and trust-aware routing using modified glowworm optimization

Improves PDR, reduces delay and energy use, isolates selfish/malicious nodes

Lacks evaluation for memory use and convergence time

Ali et al.¹⁰

CBPNN-based deep learning classifier with PSO for adaptive clustering in MANETs

Improves delivery rate and reduces drop and delay significantly

Limited real-world validation; potential computational complexity

Khedr et al.¹¹

SSA-based clustering with skyline operator for CH selection and next-hop routing

Enhances cluster stability, reduces energy consumption, and improves packet delivery and throughput

Requires computation of multiple parameters and SSA configuration; not tested in heterogeneous setups

Boualem et al.¹²

Proposed a dual-axis classification (deterministic vs. uncertainty-based) for clustering in WANETs

Enables unified understanding and systematic mapping of clustering approaches across network types

Lacks experimental performance metrics or empirical validation in real-world network scenarios

Ahmad et al.¹³

Enhanced K-means algorithm using residual energy, node density, and distance for CH selection

Improves energy balance, prolongs network lifetime, and increases delivery ratio

No evaluation under adversarial scenarios or real-world deployment constraints

Gayathiri et al.¹⁴

Proposed GKCA-LFP combines d-hop graph kernel clustering with link failure prediction (LFP) strategy.

Reduces packet loss and latency by predicting link failures and ensuring stable shortest paths.

Energy consumption metrics and adversarial resilience were not addressed in the current version.

Rathod et al.¹⁵

Hybrid routing using AODV + MBOMRP with random cryptographic fragmentation

Enhances throughput and security via multipath and hybrid encryption

Lacks evaluation on energy consumption at node-level

Aravindan et al.¹⁶

Hybrid K-Mode Clustering and Spider Monkey Optimization with Multi-Agent RL Trust Model

Enhances energy efficiency, secure routing, and packet delivery ratio

Lacks evaluation on latency and encryption time

Saravanan et al.¹⁷

Combines Modified K-means for trust-based CH selection and PE Optimization for routing

Enhances network lifetime, reduces latency, improves PDR and detection rate

Limited to simulation; real-world validation not conducted

Kumari et al.¹⁸

Used iTTM spectral clustering with GMM and CRITIC-based multi-metric CH selection.

Improves CH stability, reduces delay, enhances throughput in real urban scenarios.

High computational complexity; single-point failure due to centralized RSU approach.

Reka et al.¹⁹

Used Enhanced Chicken Swarm Optimization with APRP for CH selection and routing

Achieves high energy efficiency (91.67%), low delay (270.68 ms), high PDR (99.56%)

May face complexity in dynamic or large-scale mobility adaptation

Krishnan et al.²⁰

Self-configurable trust-based clustering with CH election via k-means and BS-verified routing

Enhances security, lowers energy consumption (to ~ 309 J), improves packet delivery (up to 61%)

May struggle with scalability and CH selection in high-density networks

Kumaresan et al.²¹

Fuzzy Marine White Shark Optimization (FMWSO) for energy-efficient CH selection in MANETs

Enhances network lifetime (5560 rounds), reduces energy (0.62 mJ)

Route maintenance is less robust under complex conditions

Panse et al.²²

Gateway node selection in heterogeneous MANETs based on cluster coverage and mobility factors

Improves routing efficiency, reduces overhead and energy use

Focused only on gateway selection; cluster head election not deeply optimized

Devi et al.²³

Bio-inspired trust-based clustering with PSO for CH selection and fuzzy logic for SCH election

Enhances energy efficiency, reduces delay, increases throughput and network lifetime

Does not explicitly address robustness to adversarial attacks or cryptographic validation

Arulprakash et al.²⁴

AVOA with Brownian motion for CH selection; MMF with SOS-based mutation for optimal routing

Enhances throughput and delivery ratio while reducing energy usage, delay, and overhead

Does not support multipath routing; bandwidth wastage due to route advertisements

Anand et al.²⁵

Dynamic Algorithm Switching with DBSCAN-based clustering

Enhances PDR, reduces delay and energy usage

No evaluation of cryptographic overheads or detailed adversarial resilience

Dilipkumar et al.²⁶

Hybrid model combining Dual Network Centrality, EGSO, and Gradient DBN

High attack detection rate, reduced memory and time overhead

Focused on DoS and Zero-Day only

Vatambeti et al.²⁷

EBDC uses star topology with energy-aware clustering and route maintenance based on link failure prediction

High accuracy (99.9%), reduced energy use, extended network lifetime

Limited to predefined topology assumptions

Hai et al.²⁸

A four-phase approach combining weight-based clustering, node-disjoint multipath routing, data offloading, and ECC-Schnorr cryptographic security

Enhances network throughput, stability, and security under attacker presence

Lacks analysis on specific energy consumption metrics and does not report memory or processing overheads

Das et al.²⁹

HGA-based CH selection with EAIICT for distributed trust and MN detection

High detection accuracy, reduced communication overhead, improved energy efficiency

Lacks explicit handling of real-time mobility models

Goswami et al.³⁰

Introduced a hybrid optimization-based routing using HS-WOA for energy-efficient MANET routing

Achieves improved throughput, lower energy consumption, reduced delay, and routing overhead

Network lifetime and robustness under security attacks not explicitly evaluated

Rajkumar et al.³¹

Integrated Enhanced Lion Swarm Optimization with ECC for CH selection and malware detection in IoT-WSNs.

Enhanced PDR (98%), energy efficiency, and robustness against sinkhole/black hole attacks.

Requires high computational overhead for ECC and adaptive LSOA in resource-constrained IoT nodes.

Sugumaran et al.³²

Zone-based clustering with ANFIS, STS data segregation, ECC encryption, IEHO

High PDR (97%), low latency, 20% energy savings, suitable for disaster use

Some metrics (e.g., trading cost, convergence time) not reported

Priya et al.³³

RL-CHS and ZBC with QR-PUF, CNN-TMS, ALEA, and HORP for MANET security routing

High PDR (99.9%), improved throughput and security via RL, CNN, and encryption

Higher energy consumption in high-density networks