Currently, hidden Markov-based multi-step attack detection models are mainly trained using the unsupervised Baum–Welch algorithm. The Baum–Welch algorithm is sensitive to the initial values of model parameters. However, its training uses random or average parameter initialization methods, which frequently results in the model training into a local optimum, thus, making the model unable to fit the alert logs well and thereby reducing the detection effectiveness of the model. To solve this issue, we propose a pre-training method for multi-step attack detection models based on the high semantic similarity of alerts in the same attack phase. The method first clusters the alerts based on their semantic information and pre-classifies the attack phase to which each alert belongs. Then, the distance of the alert vector to each attack stage is converted into the probability of generating alerts in each attack stage, replacing the initial value of Baum–Welch. The effectiveness of the proposed method is evaluated using the DARPA 2000 dataset, DEFCON21 CTF dataset, and ISCXIDS 2012 dataset. The experimental results show that the hidden Markov multi-step attack detection method based on pre-training of the proposed model parameters had higher detection accuracy than the Baum–Welch-based, K-means-based, and transfer learning differential evolution-based hidden Markov multi-step attack detection methods.

With the development of the Industrial Internet in recent years, security issues have been a hot topic of the industrial control system (ICS) network management. Identifying the protocol traffic in the communication process of the ICS is an important prerequisite to avoid security problems, especially in ICSs that use many private protocols. The private protocols cannot be analyzed due to the unknown internal structure of the protocols, which makes the ICS protocol identification work more difficult. However, the Internet-oriented protocol identification method is not applicable to the scenario of the private ICS protocols network environment. With this problem in mind, this paper proposes a method of ICS protocol identification based on the raw traffic payload. The method firstly performs data preprocessing such as data selection, interception, cleaning conversion, and labeling on the raw traffic of the protocol based on the characteristics of the industrial control protocol. Then it uses an AM-1DCNN + LSTM deep learning model to extract temporal and spatial features of the ICS raw traffic, and performs protocol identification. This method can effectively extract ICS protocol features in scenarios where protocol parsing is impossible compared with existing methods. We constructed a dataset for ICS protocol identification based on open-source data and tested the proposed method for experiments, and the identification accuracy rate reached 93%.