Table 1 Literature survey summary.

Deng et al.²⁰

Minimize the IDS reliance on data

Reduce IDS reliance on labeled data

Flow topology with convolution neural networks

Achieves high results on real-time data

Works depends on network topology

System needs to support the diverse environment

Cheng et al.²¹

Maximize the alert correlation while creating IDS systems

Alert graphical convolution networks

Maximum prediction results

Limited scalability issues

Large scale environment needs to consider.

Zhou et al.²²

Intends to solve the IDS imbalance issue

Hierarchical adversarial-attack (HAA) cohort

UNSW-SOSR2019 dataset is used to examine efficiency

Difficult to support the other dataset

Facing generalizability issues

Zeynivand et al.²⁵

Maximize the network quality

Multi-agent learning approach

High packet delivery rate and transaction rate

Scalability is difficult to manage

Create impact on network security

Roy et al.³²

Lightweight IDS for IoT

Optimization techniques (sampling, dimensionality reduction)

High detection accuracy with low false alarms

Training data dependency

Real-world deployment and dataset generalization