Table 1 Literature review.

In Ref³¹

Smart grid environment

Auto encoder-GAN

GANs excel at learning complex data distributions

Limited Scalability

HUG-RF is highly scalable, making it capable of handling large datasets in a smart grid environment.

96.4

In Ref³²

Smart grid system

Temporal convolutional network (TCN)

It enables faster computations

Inflexibility with non-temporal features

HUG-RF has flexibility in handling various data types

94.6

In Ref³³

IoT based smart grid

Decision Tree

Low Computational Cost

Decision trees are prone to overfitting

HUG-RF reduces overfitting by averaging the results of multiple trees

95.8

In Ref³⁴

Smart grid system

CNN approach

High accuracy in data pattern recognition

It is not inherently designed to optimize data transmission paths

HUG-RF optimizes data routing to ensure faster and more reliable transmission

96

In Ref³⁵

Smart grid system

DNN approach

High accuracy in complex data processing

Prone to overfitting

HUG-RF reduces overfitting by averaging the results of multiple trees

94.2

In Ref³⁶

Heterogeneous WSN routing

Improved PSO (IPSO) + Dynamic clustering

Better energy balancing, adaptive cluster head selection

Higher computation in CH selection

HUG-RF emphasizes scalability

96.8%

In Ref³⁷

Dynamic WSN routing

PSO + GA + RL + supervised learning

Improves PDR, latency, and energy efficiency

High resource usage

HUG-RF provides a lighter, ensemble-based framework

95.5%

In Ref³⁸

IoT routing in disaster-prone scenarios

Hybrid Butterfly Optimization Algorithm-PSO (BOA-PSO)

Conserves residual energy

Performance depends heavily on network configuration

HUG-RF provides more stable performance

96%