[Retracted] Software Defined Network Enabled Fog-to-Things Hybrid Deep Learning Driven Cyber Threat Detection System

Table 9

Proposed technique architecture and performance comparison with previous frameworks.

Features	Previous schemes [7–12]	Proposed technique
Architecture design	Fog nodes distributed	Centralized
DL algorithms used	1, 2, 2, 1, 2	5 models
Dataset	Anomaly based	CIDDS-01
Detection accuracy	99.82%, 98%, 99%, 98%, 92.94% in multiclass	99.92% in multiclass
Problems	Overhead, authentication issues, bad predictions, high error rate	Solved