Table 1 Comparative analysis of AI-Enhanced offloading approaches in MEC systems.

¹⁷

AI-enhanced offloading combining machine learning for Industrial IoT

Latency, Task Success Rate

Reduced latency by 23% and increased task success rate by 15%

Scalability issues in diverse industrial setups

¹⁸

Pre-trained machine learning models for service offloading in IoV

Energy Consumption, Accuracy

Energy savings of 12%, 95% task accuracy

Focused only on IoV scenarios

¹⁹

Adaptive learning in IIoT environments for offloading

Energy Efficiency, Task Completion Time

Improved energy efficiency by 17% and reduced task time by 20%

Limited evaluation of real-world IIoT systems

⁵

Multi-stage learning with predictive analytics for IoT-based smart cities

Load Management, Energy Efficiency

Achieved 30% load optimization and an 18% improvement in energy management

Limited scope to IoT-specific energy optimization

⁶

Broad Learning for Task Offloading in Industrial IoT

Adaptability, Computational Efficiency

Improved adaptability by 25% and computational efficiency by 22%

Implementation challenges in broad IoT networks

⁷

Green edge AI leveraging adaptive sensing

Energy Efficiency, Computational Accuracy

Increased energy efficiency by 19% and achieved 90% accuracy in computation

Lack of empirical real-world testing

²⁰

DRL-based collaborative networks for cloud-edge-terminal setups

Offloading Efficiency, Latency

27% better offloading efficiency, 15% reduced latency

Relies heavily on DRL model training time

Proposed work

Hybrid AI architecture combining DRL, evolutionary algorithms, and federated learning for MEC

QoE, Energy Efficiency

Up to 35% improvement in QoE and a 40% reduction in energy consumption

Requires further real-world deployment validation