Table 17 Comparative analysis of the proposed model with modern techniques.

Proposed Hybrid Model

SVM + ANN + Fuzzy Logic

Fast, interpretable, lightweight, good for small datasets

May underperform on large-scale time-series data

Well-suited for realtime smart city systems

Bilinear Spatiotemporal Fusion Network (BLSTF)

Temporal Enhancement Module,

Reduces multi-step error accumulation

Tailored for stable, periodic traffic; may underperform in highly irregular traffic patterns

Suitable for scenarios with regular traffic patterns where efficiency and interpretability are priorities¹¹

6G-Enabled Intelligent Traffic Flow Control System

Vehicle Networking, Machine Learning

Accurate traffic prediction (error rate as low as 0.81%)

Real-world implementation challenges in diverse city environments

Highly suitable for forward-looking urban traffic management systems with access to high-speed networks and smart infrastructure⁸

USTAN (Unified Spatial–Temporal Attention Network)

Spatial neighbor graph, temporal neighbor array, unified attention mechanism, gated fusion module

Captures spatial–temporal dependencies simultaneously, adaptively integrates external factors, improved prediction accuracy

May require high computational resources due to unified attention across all spatial–temporal neighbors

Moderately suitable: Offers robust spatial–temporal modeling but may not meet lightweight, real-time edge deployment constraints¹⁸

Reinforcement Learning Models

DQN, DDPG, Q-Learning

Dynamic, adaptive decision-making

Requires simulation environments and extensive tuning

Not aligned with passive traffic prediction⁹

Transformer-based Spatio-Temporal Model

Dynamic Spatial–Temporal Trend Transformer (DST2former),

Captures both dynamic and static traffic patterns effectively

Increased model complexity

Highly suitable due to its ability to model complex, non-linear spatio-temporal dependencies in realtime traffic flow data²⁴