9004 2071-8217 Спецвыпуск 2023 1-194

RAR RUS 13-24 0009-0006-6856-2108 Peter the Great St. Petersburg Polytechnic University Pahomov Maksim pahomov_ma@spbstu.ru 0000-0003-1345-1874 Peter the Great St. Petersburg Polytechnic University Pavlenko Evgeny pavlenko_eyu@spbstu.ru 0009-0008-8784-8720 Peter the Great St. Petersburg Polytechnic University Sobolev Nikolay sobolev.nv@edu.spbstu.ru Analysis of methods for ensuring information security of wireless AD-HOC networks The features of wireless self-organizing networks and their routing mechanisms are analyzed. The classification of attacks on this type of networks is presented. Groups of methods used to ensure the security of self-organizing networks are highlighted. The analysis of representatives of each group of methods is performed, their advantages and disadvantages are singled out. The purpose and direction of further research is formulated. 004.056 information security wireless ad-hoc networks https://jisp.spbstu.ru/article/2023.12.1/ 2023_spetsvipusk_ru_.pdf

RAR RUS 25-33 0000-0002-2141-6780 Peter the Great Saint-Petersburg Polytechnic University Shtyrkina Anna anna_sh@ibks.spbstu.ru 0000-0002-7485-4848 Peter the Great St. Petersburg Polytechnic University Aleksandrova Elena aleksandrova_eb@spbstu.ru Cyber physical attacks detection based on graph Fourier transform The paper is devoted to the topic of detecting cyber-physical systems (CPS) attacks that affect the parameters of the functioning of devices. The potential consequences of cyber attacks on the CPS, as well as the corresponding changes in the modeling graph, are considered. A method for detecting cyberattacks based on the graph Fourier transform and the gradient boosting algorithm is proposed. The method makes it possible to detect a non-standard change in the operation parameters of devices and evaluate its criticality from the point of view of the centrality of a group of modeling vertices. 004.056 cyber-physical systems graph theory graph signal processing boosting https://jisp.spbstu.ru/article/2023.12.2/ 2023_spetsvipusk_ru_.pdf

RAR RUS 34-44 0000-0002-6370-123X Peter the Great St. Petersburg Polytechnic University Danilov Vladislav danilov.wrk@gmail.com Peter the Great St. Petersburg Polytechnic University Gribkov Nikita gribkov.na@edu.spbstu.ru 0000-0001-8206-2915 Peter the Great St. Petersburg Polytechnic University Ivanov Denis ivanov@ibks.spbstu.ru Analysis of methods for detecting artificially synthesized content The paper presents an analysis of existing methods for detecting artificially synthesized content and proposes a proprietary architecture for DeepFake’s hybrid detection system based on searching original content. The study tests and compares the effectiveness of detection methods in two different cases. In the first case, records for training and testing samples are used from the same dataset; in the second case, testing is performed using a black-box method using records from different datasets. As a result, it is concluded that there are shortcomings in the existing methods and a hybrid DeepFake detection system architecture is proposed. 004.56 DeepFake detection generative adversarial networks artificially synthesized content deep learning https://jisp.spbstu.ru/article/2023.12.3/ 2023_spetsvipusk_ru_.pdf

RAR RUS 45-53 Peter the Great St. Petersburg Polytechnic University Zagalsky Dmitry dzagalskii@yandex.ru Peter the Great St. Petersburg Polytechnic University Kashkarov Oleg Peter the Great St. Petersburg Polytechnic University Moskvin Dmitry moskvin_da@spbstu.ru Peter the Great St. Petersburg Polytechnic University Solovey Roman Peter the Great St. Petersburg Polytechnic University LOGINOV Zakhar loginoff.zahar@yandex.ru Development of a metaverse model for user access control to metaverse resources In this article, various approaches to the definition of the concept of the metaverse are investigated, and a proper, most universal one is proposed. The threats are analyzed and the requirements for the information security of the metaverse are highlighted. Its main components and features of their use are investigated. The applicability of various access control models to ensure effective security management of the metaverse has been evaluated. 004.056 metaverse metaverse security metaverse model https://jisp.spbstu.ru/article/2023.12.4/ 2023_spetsvipusk_ru_.pdf

RAR RUS 54-64 0009-0005-3102-5950 Peter the Great St. Petersburg Polytechnic University Izotova Oksana izotova@ibks.spbstu.ru 0000-0003-2849-4682 Peter the Great St. Petersburg Polytechnic University Lavrova Daria lavrova_ds@spbstu.ru

Russia, 195251, St. Petersburg, Polytechnicheskaya str., 29

Early detection of network attacks based on weight agnostic neural networks This paper describes an approach to early detection of network attacks using weight agnostic neural networks. The choice of the type of neural networks is due to the specificity of their architecture that provides high processing speed and performance, which is significant in solving the problem of early attack detection. Experimental studies have demonstrated the effectiveness of the proposed approach based on a combination of multiple regression for feature selection of the training sample and weight agnostic neural networks. The accuracy of attack detection is comparable to the best results in the field with a significant time gain. 004.056 network attacks weight agnostic neural networks multiple regression machine learning https://jisp.spbstu.ru/article/2023.12.5/ 2023_spetsvipusk_ru_.pdf

RAR RUS 65-76 Peter the Great St. Petersburg Polytechnic University Berko Alexander 0000-0001-9659-1244 Peter the Great St. Petersburg Polytechnic University Poltavtseva Maria potavtseva@ibks.spbstu.ru A framework for security policies modeling for big data systems This paper is about the task of automating the analysis of access control in big data management systems by modeling security policies. The paper analyzes modern methods of access control in this class of systems, defines the requirements and selects the most promising one to describe the security policy within the framework of the developed solution. The task of modeling security policies in big data management systems is set. The architecture, main components and generalized algorithm of the software framework for its solution are presented. The results of experimental validation are also presented, the advantages and disadvantages of the framework are evaluated and the ways of its further development are proposed. 004.056 big data security big data management systems access control attribute access control security policy security policy modeling https://jisp.spbstu.ru/article/2023.12.6/ 2023_spetsvipusk_ru_.pdf

RAR RUS 77-85 0009-0008-7164-1872 Peter the Great St. Petersburg Polytechnic University Skokov Nikita vampire@ns-skokov.ru 0000-0002-2264-7513 Peter the Great St. Petersburg Polytechnic University Krundyshev Vasiliy krundyshev_vm@spbstu.ru Ensuring the security of web resources based on intelligent analysis of network traffic This paper presents the developed method for detecting anomalies in network traffic, which is based on the technology of hierarchical temporary memory. To evaluate the effectiveness of the proposed solution, a new data set was generated containing information about legitimate and malicious network sessions. As a result of experimental studies, it was found that the use of a hierarchy of features and support for the memory mechanism make it possible to reveal hidden patterns in the analyzed chains of network requests to web resources with high accuracy. 004.056 network traffic analysis web resource security hierarchical temporary memory anomaly detection network attacks https://jisp.spbstu.ru/article/2023.12.7/ 2023_spetsvipusk_ru_.pdf

RAR RUS 86-94 0000-0003-0623-9891 Peter the Great St. Petersburg Polytechnic University Gololobov Nikita gololobov_nv@spbstu.ru 0000-0003-1345-1874 Peter the Great St. Petersburg Polytechnic University Pavlenko Evgeny pavlenko_eyu@spbstu.ru Classification of methods to counteract a data poisoning attack during neural network training Considered methods of counteraction attack data poisoning type learning neural network and compiled a model of the attacker, according to which a classification of the considered methods. The classification obtained as a result of the study can be used in further research, the ultimate goal of which is to increase the level of unification and automation of data processing and protection methods. 004.056 information security machine learning data processing data poisoning learning quality https://jisp.spbstu.ru/article/2023.12.8/ 2023_spetsvipusk_ru_.pdf

RAR RUS 95-107 Peter the Great St. Petersburg Polytechnic University Soshnev Maxim 0000-0002-9732-0099 Peter the Great St. Petersburg Polytechnic University Kalinin Maxim max@ibks.spbstu.ru

Russia, 195251, St. Petersburg, Polytechnicheskaya str., 29

Protection against the threat of the machine learning models extraction The threat of extraction of the machine learning models is considered. Most of the modern approaches to the prevention of machine learning models extraction are based on the use of the protective noising mechanism. The main disadvantage of this protective method is the decrease in the accuracy of the outputs generated by the protected model. The paper states the requirements for methods for protecting machine learning models against extraction and presents a new method, which supplements noise with a distillation stage. It has been experimentally shown that the developed method ensures the resistance of machine learning models to extraction while maintaining the quality of their results by transforming the protected models to other, the simplified, but equivalent, models. 004.056 machine learning security model distillation noising soft label degree of security accuracy of results model extraction threat https://jisp.spbstu.ru/article/2023.12.9/ 2023_spetsvipusk_ru_.pdf

RAR RUS 108-119 Peter the Great St. Petersburg Polytechnic University Rudnitskaya Ekaterina 0000-0001-9659-1244 Peter the Great St. Petersburg Polytechnic University Poltavtseva Maria potavtseva@ibks.spbstu.ru Privacy of machine learning models The paper is devoted to the problem of ensuring the confidentiality of models in machine learning systems. The aim of the work is to ensure the confidentiality of proprietary models of machine learning systems. In the course of the work we analyzed attacks aimed at violating the confidentiality of models of machine learning systems, as well as ways to protect against this type of attacks, as a result of which the problem of protection against such attacks is set as a search for anomalies in the input data. We propose a way to detect anomalies in the input data based on statistical data, taking into account the resumption of the attack under a different account of the attacker. The obtained results can be used as a basis for designing components of machine learning defense systems. 004.056 information security artificial intelligence artificial intelligence security attacks on machine learning systems privacy model privacy behavioral analysis https://jisp.spbstu.ru/article/2023.12.10/ 2023_spetsvipusk_ru_.pdf

RAR RUS 120-129 0000-0003-0623-9891 Peter the Great St. Petersburg Polytechnic University Gololobov Nikita gololobov_nv@spbstu.ru 13103571000 0000-0002-0232-7248 Peter the Great St. Petersburg Polytechnic University Zegzhda Dmitry zegzhda_dp@spbstu.ru

Russia, 195251, St. Petersburg, Polytechnicheskaya str., 29

A set-theoretic model of data poisoning attack techniques in artificial intelligence systems The article contains formalized techniques of data poisoning attacks are presented in the form of a set-theoretic model, considering the levels at which poisoning can be carried out. The division of attacks according to levels allows further consideration of each type of poisoning attack to prevent or minimize the consequences of data contamination specific to each level. The model obtained because of the study can be used in further research, the goal of which is to increase the level of unification and automation of data processing and protection methods. 004.056 cybersecurity data poisoning data cleaning heterogeneous data set-theoretic attack model https://jisp.spbstu.ru/article/2023.12.11/ 2023_spetsvipusk_ru_.pdf

RAR RUS 130-138 0000-0001-9862-1507 Peter the Great St. Petersburg Polytechnic University Dakhnovich Andrey add@ibks.spbstu.ru Peter the Great St. Petersburg Polytechnic University Solovey Roman Peter the Great St. Petersburg Polytechnic University LOGINOV Zakhar loginoff.zahar@yandex.ru Peter the Great St. Petersburg Polytechnic University Zagalsky Dmitry dzagalskii@yandex.ru Peter the Great St. Petersburg Polytechnic University Moskvin Dmitry moskvin_da@spbstu.ru 13103571000 0000-0002-0232-7248 Peter the Great St. Petersburg Polytechnic University Zegzhda Dmitry zegzhda_dp@spbstu.ru

Russia, 195251, St. Petersburg, Polytechnicheskaya str., 29

Caviats of detecting unfairness biases in results of recommender systems In the context of the deep penetration of information technologies and services into people’s lives, the issues of control over recommendation systems (hereinafter – RS), which are actively used by social networks and Internet applications for personalized selection and ranking of content for users, are becoming increasingly relevant. The concept of RS operation is based on the preliminary collection of various types and degrees of sensitivity data about the user and their subsequent algorithmic processing in order to provide personalized recommendations. Personalized recommendations selected according to certain methods can create different worldviews for the same users, provoke active actions, etc. Thus, there is a need for a tool to assess the susceptibility of RS to the influences that lead to the bias of recommendation algorithms, on behalf of an external observer. 004.056 recommender systems unfairness biases social networks social network communications cybersecurity https://jisp.spbstu.ru/article/2023.12.12/ 2023_spetsvipusk_ru_.pdf

RAR RUS 139-144 0009-0004-1271-709X Peter the Great St. Petersburg Polytechnic University Ognev Roman ognev_ra@spbstu.ru 13103571000 0000-0002-0232-7248 Peter the Great St. Petersburg Polytechnic University Zegzhda Dmitry zegzhda_dp@spbstu.ru

Russia, 195251, St. Petersburg, Polytechnicheskaya str., 29

An empirical study of the robustness of a linear filter built on the Neyman – Pearson criterion to a change in the mean values The assertion about the stability of a linear filter built on the basis of the Neyman – Pearson criterion was verified by performing falsifying experiments. The relationship between the eigenvalues of the interference covariance matrix and their minimum values and network stability was not found. 004.56 linear filter single-layer neural network robustness Neyman – Pearson criterion https://jisp.spbstu.ru/article/2023.12.13/ 2023_spetsvipusk_ru_.pdf

RAR RUS 145-158 0009-0008-4442-5365 Peter the Great St. Petersburg Polytechnic University Kubrin Georgiy kubrin@ibks.spbstu.ru 13103571000 0000-0002-0232-7248 Peter the Great St. Petersburg Polytechnic University Zegzhda Dmitry zegzhda_dp@spbstu.ru

Russia, 195251, St. Petersburg, Polytechnicheskaya str., 29

Vulnerability detection with an ensemble of analysis algorithms for code graph representation The paper presents an analysis of existing methods for software vulnerabilities detection. A problem of faulty paths in interprocedural code graph representation is presented. This problem hinders application of graph deep learning models to code analysis tasks. A method based on an ensemble of algorithms for code graph analysis is presented to overcome the problem of faulty paths. The method performs gradual reduction of analyzed code fragments size for effective application of algorithms with high time complexity. A prototype of vulnerability detection system for .NET software based on the proposed method is presented. The prototype is evaluated using NIST SARD database and software with considerable codebase size. 004.56 software vulnerabilities detection logical vulnerabilities static code analysis graph theory deep learning https://jisp.spbstu.ru/article/2023.12.14/ 2023_spetsvipusk_ru_.pdf

RAR RUS 159-172 0000-0001-9665-0128 Peter the Great St. Petersburg Polytechnic University Suprun Alexander afs54@inbox.ru

Russia, 195251, St. Petersburg, Polytechnicheskaya str., 29

0009-0008-8784-8720 Peter the Great St. Petersburg Polytechnic University Sobolev Nikolay sobolev.nv@edu.spbstu.ru A method for detecting attacks on web applications using a web application firewall based on an artificial neural network A method for detecting network attacks on web applications using a neural network based on LSTM is presented. The process of extracting the necessary information from traffic before submitting it to an artificial neural network (ANN) is presented. This process of preprocessing HTTP traffic allows you to select key fragments, which are subsequently vectorized for proper processing in the ANN. The ANN architecture is defined, including the necessary layers, for the multiclassification of HTTP traffic and the detection of network attacks on web applications. 004.056 firewall web-applications network attacks https://jisp.spbstu.ru/article/2023.12.15/ 2023_spetsvipusk_ru_.pdf

RAR RUS 173-182 Peter the Great St. Petersburg Polytechnic University Kuzmina Kseniia 0000-0002-7121-6031 Peter the Great St. Petersburg Polytechnic University Yarmak Anastasiia yarmak_av@spbstu.ru 0000-0002-7485-4848 Peter the Great St. Petersburg Polytechnic University Aleksandrova Elena aleksandrova_eb@spbstu.ru Proxy signature based on GOST 34.10–2018 The paper proposes a digital signature scheme that expands the functionality of the GOST 34.10–2018 and allows delegating signing capability to a trusted person (proxy signer). A classification of proxy signatures has been developed; the selected delegation scheme was modified to prevent misuse by a proxy signer. The correctness of the scheme was shown, the analysis of compliance with security requirements was carried out. The results of software implementation testing are presented. 004.056 digital signature proxy signature GOST 34.10–2018 elliptic curves https://jisp.spbstu.ru/article/2023.12.16/ 2023_spetsvipusk_ru_.pdf

RAR RUS 183-193 0009-0000-3181-4769 Peter the Great St. Petersburg Polytechnic University Kostin Sergey s8kostin@gmail.com 0000-0002-7485-4848 Peter the Great St. Petersburg Polytechnic University Aleksandrova Elena aleksandrova_eb@spbstu.ru Research of the isogeny graph structure of supersingular curves for post-quantum cryptography protocols Isogeny graphs of supersingular elliptic curves are one of perspective mathematical structures for post-quantum cryptography algorithms. However, recently published attack on the SIDH protocol [1] demonstrates that isogeny graphs require a more detailed study when they are used in real protocols. In this paper, we analyze the structure of isogeny graphs of degree l > 2 and consider a set of nodes of a special kind to which the attack [7] on path recovery in the graph is applicable. 004.056 post-quantum cryptography isogeny graphs https://jisp.spbstu.ru/article/2023.12.17/ 2023_spetsvipusk_ru_.pdf