Malware has emerged as a significant threat with growing infection rates
and degrees of sophistication as the number of devices and technologies related to the
Internet of Things (IoT) has increased and more are put into service. Without robust
security procedures, many confidential records are left open to vulnerabilities. As a
result, it is simple for cybercriminals to use this data to carry out various unlawful acts.
Therefore, advanced network security mechanisms that are capable of executing a
traffic analysis in real time and mitigating harmful traffic are required. These
mechanisms must also be able to detect malicious traffic. We propose a revolutionary
technique for IoT malware traffic analysis that uses deep learning and graphical
demonstration to detect and categorize new malware more quickly. This will allow us
to handle the difficulty that has been presented (zero-day malware). Due to the
utilization of deep learning technology, the suggested method for detecting malicious
network traffic operates at the package level, significantly reducing the time required
for detection and producing promising outcomes. A dataset called “1000 pcap files of
ordinary and malicious traffic that were collected from various network traffic sources”
is created to evaluate our method's performance. This dataset is used to assess how well
our method works. The experimental findings of the Residual Neural Network
(ResNet) are highly encouraging, delivering a rate of accuracy for the identification of
malicious traffic that is 95.09%.
Keywords: Intrusion detection system (IDS), Machine Learning (ML), Network, Security, Traffic.
