Table 5 Comprehensive Comparative Analysis of Secure and Learning-Based Routing Protocols for Zone-Based MANETs.
Protocol | Author (Year) | Zone-Based | Learning Type | Security Mechanism | Energy Eff. | Limitation | Overhead | Convergence Speed | Anomaly Detection | Resource Cost |
|---|---|---|---|---|---|---|---|---|---|---|
RLSRP (Proposed) | — | Yes (k-hop) | Deep Q-Learning | Latency-based wormhole detection | High | Needs parameter tuning | \(\mathscr {O}(k \log n)\) | Fast (via DQN) | RTT deviation monitoring | Low |
Cluster-RL | Zhang et al. (2022)29 | Yes | Tabular Q-Learning | None | Low | No security; not attack-adaptive | \(\mathscr {O}(n^2)\) | Moderate | Reputation feedback-based | Moderate |
FSSAM | Rath et al. (2020)27 | No | None | Fuzzy trust model | Moderate | High computation; poor scalability | \(\mathscr {O}(n)\) | Slow | Trust deviation detection | High |
Reputation Q-Learning | Chen et al. (2021)28 | No | Q-Learning | Reputation-based trust system | Low | Static trust; lacks zone awareness | \(\mathscr {O}(n \log n)\) | Slow | Behavioural monitoring | Moderate |
ZRDM+LFPM | K et al. (2021)33 | Yes | None | None | Moderate | No learning; no attack detection | \(\mathscr {O}(n)\) | Slow | None | Moderate |
KB Adaptive Routing | Kavitha et al. (2022)34 | No | Rule-based | Rule-based detection | Moderate | No clustering; fixed detection logic | Rule-based | Moderate | Signature-based | Moderate |
ECC Routing | Shukla et al. (2021)35 | No | None | ECC-based encryption | Moderate | High cryptographic cost | High | Moderate | None | High |
HMM Defense | Kalkha et al. (2019)36 | No | Statistical HMM | HMM-based anomaly detection | Low | No energy model; poor scalability | Statistical | Slow | HMM sequence deviations | Low |
DAPV | Li et al. (2019)37 | No | Anomaly Detection | Provenance + verification | Low | Provenance verification overhead | \(\mathscr {O}(n)\) | Moderate | Provenance chain mismatch | Low |
Mod. Sec. AODV | Narayanan & Murugaboopathi (2020)38 | No | None | Wormhole blocking | Moderate | No learning; reactive only | \(\mathscr {O}(n)\) | Fast | Link-level blocking | Moderate |
Bee Trust AODV | Keerthika & Malarvizhi (2019)39 | No | Bio-inspired | Bee-trust model | Low | No zone awareness; no deep learning | Bio-inspired | Slow | Bee colony optimisation trust | Low |
