Industrial Control Systems (ICS) are evolving into smart environments with increased interconnectivity by being connected to the Internet. These changes increase the likelihood of security vulnerabilities and accidents. As the risk of cyberattacks on ICS has increased, various anomaly detection studies are being conducted to detect abnormal situations in industrial processes. However, anomaly detection in ICS suffers from numerous false alarms. When false alarms occur, multiple sensors need to be checked, which is impractical. In this study, when an anomaly is detected, sensors displaying abnormal behavior are visually presented through XAI-based analysis to support quick practical actions and operations. Anomaly Detection has designed and applied better anomaly detection technology than the first prize at HAICon2020, an ICS security threat detection AI contest hosted by the National Security Research Institute last year, and explains the anomalies detected in its model. To the best of our knowledge, our work is at the forefront of explainable anomaly detection research in ICS. Therefore, it is expected to increase the utilization of anomaly detection technology in ICS.

According to a study by Cybersecurity Ventures, cybercrime is expected to cost $6 trillion annually by 2021. Most cybersecurity threats access internal networks through infected endpoints. Recently, various endpoint environments such as smartphones, tablets, and Internet of things (IoT) devices have been configured, and security issues caused by malware targeting them are intensifying. Event logs-based detection technology for endpoint security is detected using rules or patterns. Therefore, known attacks can respond, but unknown attacks can be difficult to respond to immediately. To solve this problem, in this paper, local outlier factor (LOF) and Autoencoder detect suspicious behavior that deviates from normal behavior. It also detects threats and shows the corresponding threats when suspicious events corresponding to the rules created through the attack profile are constantly occurring. Experimental results detected eight new suspicious processes that were not previously detected, and four malicious processes and one suspicious process were judged using Hybrid Analysis and VirusTotal. Based on the experiment results, it is expected that the use of operational policies such as allowlists in the proposed model will significantly improve performance by minimizing false positives.

Most cyberattacks use malicious codes, and according to AV-TEST, more than 1 billion malicious codes are expected to emerge in 2020. Although such malicious codes have been widely seen around the PC environment, they have been on the rise recently, focusing on IoT devices such as smartphones, refrigerators, irons, and various sensors. As is known, Linux/embedded environments support various architectures, so it is difficult to identify the architecture in which malware operates when analyzing malware. This paper proposes an AI-based malware analysis technology that is not affected by the operating system or architecture platform. The proposed technology works intuitively. It uses platform-independent binary data rather than features based on the structured format of the executable files. We analyzed the strings from binary data to classify malware. The experimental results achieved 94% accuracy on Windows and Linux datasets. Based on this, we expect the proposed technology to work effectively on other platforms and improve through continuous operation/verification.

Many previous studies have investigated applying artificial intelligence (AI) to cyber security. Despite considerable performance advantages, AI for cyber security requires final confirmation by an analyst, e.g. malware misdetection can cause significant adverse side effects. Thus, a human analyst must check all AI predictions, which poses a major obstacle to AI expansion. This paper proposes a reliability indicator for AI prediction using explainable artificial intelligence and statistical analysis techniques. This will enable analysts with limited daily workload to focus upon valuable data, and quickly verify AI predictions. Analysts generally make decisions based on several features that they know exactly what they mean, rather than all available features. Since the proposed reliability indicator is calculated using features meaningful to analysts, it can be easily understood and hence speed final decisions. To verify the performance of the proposed method, an experiment was conducted using the IDS dataset and the malware dataset. The AI error was detected better than the existing AI model at about 114% in IDS and 95% in malware. Thus, cyberattack response could be greatly improved by adopting the proposed method.

Malicious codes, such as advanced persistent threat (APT) attacks, do not operate immediately after infecting the system, but after receiving commands from the attacker’s command and control (C&C) server. The system infected by the malicious code tries to communicate with the C&C server through the IP address or domain address of the C&C server. If the IP address or domain address is hard-coded inside the malicious code, it can analyze the malicious code to obtain the address and block access to the C&C server through security policy. In order to circumvent this address blocking technique, domain generation algorithms are included in the malware to dynamically generate domain addresses. The domain generation algorithm (DGA) generates domains randomly, so it is very difficult to identify and block malicious domains. Therefore, this paper effectively detects and classifies unknown DGA domains. We extract features that are effective for TextCNN-based label prediction, and add additional domain knowledge-based features to improve our model for detecting and classifying DGA-generated malicious domains. The proposed model achieved 99.19% accuracy for DGA classification and 88.77% accuracy for DGA class classification. We expect that the proposed model can be applied to effectively detect and block DGA-generated domains.