Все выпуски

This paper is devoted to the problem of edge criticality identification and ranking in complex networks, which is a part of a modern research direction in the novel network science. The diffusion importance belongs to the set of acknowledged methods that help to identify the significant connections in the graph that are critical to retaining structural integrity. In the present work, we develop the Iterative Diffusion Importance algorithm that is based on the re-estimation of critical topological features at each step of the graph deconstruction. The Iterative Diffusion Importance has been compared with methods such as diffusion importance and degree product, which are two very well-known benchmark algorithms. As for benchmark networks, we tested the Iterative Diffusion Importance on three standard networks, such as Zachary’s Karate Club, the American Football Network, and the Dolphins Network, which are often used for algorithm efficiency evaluation and are different in size and density. Also, we proposed a new benchmark network representing the airplane communication between Japan and the US. The numerical experiment on finding the ranking of critical edges and the following network decomposition demonstrated that the proposed Iterative Diffusion Importance exceeds the conventional diffusion importance by the efficiency for 2–35% depending on the network complexity, the number of nodes, and the number of edges. The only drawback of the Iterative Diffusion Importance is an increase in computation complexity and hencely in the runtime, but this drawback can be easily compensated for by the preliminary planning of the network deconstruction or protection and by reducing the re-evaluation frequency of the iterative process.

This paper is devoted to the problem of edge criticality identification and ranking in complex networks, which is a part of a modern research direction in the novel network science. The diffusion importance belongs to the set of acknowledged methods that help to identify the significant connections in the graph that are critical to retaining structural integrity. In the present work, we develop the Iterative Diffusion Importance algorithm that is based on the re-estimation of critical topological features at each step of the graph deconstruction. The Iterative Diffusion Importance has been compared with methods such as diffusion importance and degree product, which are two very well-known benchmark algorithms. As for benchmark networks, we tested the Iterative Diffusion Importance on three standard networks, such as Zachary’s Karate Club, the American Football Network, and the Dolphins Network, which are often used for algorithm efficiency evaluation and are different in size and density. Also, we proposed a new benchmark network representing the airplane communication between Japan and the US. The numerical experiment on finding the ranking of critical edges and the following network decomposition demonstrated that the proposed Iterative Diffusion Importance exceeds the conventional diffusion importance by the efficiency for 2–35% depending on the network complexity, the number of nodes, and the number of edges. The only drawback of the Iterative Diffusion Importance is an increase in computation complexity and hencely in the runtime, but this drawback can be easily compensated for by the preliminary planning of the network deconstruction or protection and by reducing the re-evaluation frequency of the iterative process.

В работе исследуется поиск неэффективных ребер в модели стабильной динамики Нестрова–де Пальмы (2003). Для этой цели мы доказываем несколько общих теорем о свойствах равновесия, в том числе о том, что условие равенства стоимостей для всех используемых маршрутов может быть распространено на все пути, задействующие ребра из равновесных маршрутов. В работе показывается, что стандартная постановка задачи о поиске ребер, удаление которых приводит к уменьшению стоимости проезда для всех участников, не имеет практического смысла, так как одно и то же ребро может быть как эффективным, так и неэффективным (в зависимости от загрузки сети). В работе мы вводим понятие неэффективного ребра, опираясь на чувствительность суммарных издержек водителей к издержкам на ребре. В работе приводятся алгоритм поиска неэффективных ребер и результаты численных экспериментов для транспортной сети города Анахайм.

The work investigates the search for inefficient edges in the model of stable dynamics by Nestrov – de Palma (2003). For this purpose, we prove several general theorems about equilibrium properties, including the condition of equal costs for all used routes that can be extended to all paths involving edges from equilibrium routes. The study demonstrates that the standard problem formulation of finding edges whose removal reduces the cost of travel for all participants has no practical significance because the same edge can be both efficient and inefficient depending on the network’s load. In the work, we introduce the concept of an inefficient edge based on the sensitivity of total driver costs to the costs on the edge. The paper provides an algorithm for finding inefficient edges and presents the results of numerical experiments for the transportation network of the city of Anaheim.

The Industrial Internet of Things (IIoT) networks plays a significant role in enhancing industrial automation systems by connecting industrial devices for real time data monitoring and predictive maintenance. However, this connectivity introduces new vulnerabilities which demand the development of advanced intrusion detection systems. The nuclear facilities are considered one of the closest examples of critical infrastructures that suffer from high vulnerability through the connectivity of IIoT networks. This paper develops a robust intrusion detection approach using machine and ensemble learning algorithms specifically determined for IIoT networks. This approach can achieve optimal performance with low time complexity suitable for real-time IIoT networks. For each algorithm, Grid Search is determined to fine-tune the hyperparameters for optimizing the performance while ensuring time computational efficiency. The proposed approach is investigated on recent IIoT intrusion detection datasets, WUSTL-IIOT-2021 and Edge-IIoT-2022 to cover a wider range of attacks with high precision and minimum false alarms. The study provides the effectiveness of ten machine and ensemble learning models on selected features of the datasets. Synthetic Minority Over-sampling Technique (SMOTE)-based multi-class balancing is used to manipulate dataset imbalances. The ensemble voting classifier is used to combine the best models with the best hyperparameters for raising their advantages to improve the performance with the least time complexity. The machine and ensemble learning algorithms are evaluated based on accuracy, precision, recall, F1 Score, and time complexity. This evaluation can discriminate the most suitable candidates for further optimization. The proposed approach is called the XCL approach that is based on Extreme Gradient Boosting (XGBoost), CatBoost (Categorical Boosting), and Light Gradient- Boosting Machine (LightGBM). It achieves high accuracy, lower false positive rate, and efficient time complexity. The results refer to the importance of ensemble strategies, algorithm selection, and hyperparameter optimization in enhancing the performance to detect the different intrusions across the IIoT datasets over the other models. The developed approach produced a higher accuracy of 99.99% on the WUSTL-IIOT-2021 dataset and 100% on the Edge-IIoTset dataset. Our experimental evaluations have been extended to the CIC-IDS-2017 dataset. These additional evaluations not only highlight the applicability of the XCL approach on a wide spectrum of intrusion detection scenarios but also confirm its scalability and effectiveness in real-world complex network environments.

The Industrial Internet of Things (IIoT) networks plays a significant role in enhancing industrial automation systems by connecting industrial devices for real time data monitoring and predictive maintenance. However, this connectivity introduces new vulnerabilities which demand the development of advanced intrusion detection systems. The nuclear facilities are considered one of the closest examples of critical infrastructures that suffer from high vulnerability through the connectivity of IIoT networks. This paper develops a robust intrusion detection approach using machine and ensemble learning algorithms specifically determined for IIoT networks. This approach can achieve optimal performance with low time complexity suitable for real-time IIoT networks. For each algorithm, Grid Search is determined to fine-tune the hyperparameters for optimizing the performance while ensuring time computational efficiency. The proposed approach is investigated on recent IIoT intrusion detection datasets, WUSTL-IIOT-2021 and Edge-IIoT-2022 to cover a wider range of attacks with high precision and minimum false alarms. The study provides the effectiveness of ten machine and ensemble learning models on selected features of the datasets. Synthetic Minority Over-sampling Technique (SMOTE)-based multi-class balancing is used to manipulate dataset imbalances. The ensemble voting classifier is used to combine the best models with the best hyperparameters for raising their advantages to improve the performance with the least time complexity. The machine and ensemble learning algorithms are evaluated based on accuracy, precision, recall, F1 Score, and time complexity. This evaluation can discriminate the most suitable candidates for further optimization. The proposed approach is called the XCL approach that is based on Extreme Gradient Boosting (XGBoost), CatBoost (Categorical Boosting), and Light Gradient- Boosting Machine (LightGBM). It achieves high accuracy, lower false positive rate, and efficient time complexity. The results refer to the importance of ensemble strategies, algorithm selection, and hyperparameter optimization in enhancing the performance to detect the different intrusions across the IIoT datasets over the other models. The developed approach produced a higher accuracy of 99.99% on the WUSTL-IIOT-2021 dataset and 100% on the Edge-IIoTset dataset. Our experimental evaluations have been extended to the CIC-IDS-2017 dataset. These additional evaluations not only highlight the applicability of the XCL approach on a wide spectrum of intrusion detection scenarios but also confirm its scalability and effectiveness in real-world complex network environments.

В настоящее время интенсивное развитие получило направление в рамках теории распределенных вычислений, когда вычислительные задачи решаются распределенно коллективом ресурсно ограниченных устройств. На практике такой сценарий реализуется при обработке данных в системах интернета вещей, когда с целью снижения латентности систем и загруженности сетевой инфраструктуры данные обрабатываются на вычислительных устройствах края сети, в то время как стремительный рост и распространение систем интернета вещей ставят вопрос о необходимости разработки методов снижения ресурсоемкости производимых вычислений. Ресурсная ограниченность вычислительных устройств ставит следующие вопросы распределения вычислительных ресурсов: во-первых, необходимость учета ресурсной стоимости транзита данных между решаемыми на различных устройствах задачах, во-вторых, необходимость учета ресурсной стоимости непосредственно процесса распределения вычислительных ресурсов, что особенно актуально для групп автономных устройств (роботы различных типов, сенсорные сети и др.). Анализ современных публикаций, представленных в открытом доступе, продемонстрировал отсутствие предложенных моделей или методов распределения вычислительных ресурсов, которые бы совместно учитывали перечисленное, что делает создание новой математической модели организации распределенных вычислений в системах интернета вещей и методов ее решения актуальными.

В данной статье предложены новая математическая модель распределения вычислительных ресурсов и эвристические методы решения получаемой задачи оптимизации, что в комплексе реализует организацию распределенных вычислений в системах интернета вещей. Рассматривается сценарий, когда в группе устройств имеется лидер, который принимает решение о распределении вычислительных ресурсов, в том числе и собственных, для распределенного решения вычислительных задач с наличием информационных обменов. Также предполагается, что отсутствует априорная информация о том, какому устройству назначена роль лидера, и о маршрутах миграции вычислительных задач на устройства.

Результаты экспериментального исследования продемонстрировали целесообразность использования предложенных моделей и эвристических методов: достигается распределение вычислительных ресурсов со снижением ресурсной стоимости решения вычислительной задачи до 52 % при учете ресурсной стоимости транзита данных, экономия ресурсов до 73 % при дополнении основных критериев оптимизации распределения задач критерием минимизации количества и расстояний миграций подзадач вычислительной задачи (ВЗ), а также снижение ресурсной стоимости решения задачи распределения вычислительных ресурсов до 28 раз со снижением качества полученного распределения до 10 %.

Currently, a significant development has been observed in the direction of distributed computing theory, where computational tasks are solved collectively by resource-constrained devices. In practice, this scenario is implemented when processing data in Internet of Things systems, with the aim of reducing system latency and network infrastructure load, as data is processed on edge network computing devices. However, the rapid growth and widespread adoption of IoT systems raise questions about the need to develop methods for reducing the resource intensity of computations. The resource constraints of computing devices pose the following issues regarding the distribution of computational resources: firstly, the necessity to account for the transit cost between different devices solving various tasks; secondly, the necessity to consider the resource cost associated directly with the process of distributing computational resources, which is particularly relevant for groups of autonomous devices such as drones or robots. An analysis of modern publications available in open access demonstrated the absence of proposed models or methods for distributing computational resources that would simultaneously take into account all these factors, making the creation of a new mathematical model for organizing distributed computing in IoT systems and its solution methods topical. This article proposes a novel mathematical model for distributing computational resources along with heuristic optimization methods, providing an integrated approach to implementing distributed computing in IoT systems. A scenario is considered where there exists a leader device within a group that makes decisions concerning the allocation of computational resources, including its own, for distributed task resolution involving information exchanges. It is also assumed that no prior knowledge exists regarding which device will assume the role of leader or the migration paths of computational tasks across devices. Experimental results have shown the effectiveness of using the proposed models and heuristics: achieving up to a 52% reduction in resource costs for solving computational problems while accounting for data transit costs, saving up to 73% of resources through supplementary criteria optimizing task distribution based on minimizing fragment migrations and distances, and decreasing the resource cost of resolving the computational resource distribution problem by up to 28 times with reductions in distribution quality up to 10%.

При обследовании методом однофотонной эмиссионной компьютерной томографии (ОФЭКТ) пациентам с заболеваниями костной системы вводится радиофармпрепарат (РФП), который специфическим образом накапливается в патологических очагах. Количественные оценки накопления РФП в очагах важны для определения стадии заболевания, прогнозирования его течения и разработки персонализированных терапевтических стратегий. Исследования точности количественных оценок обычно проводятся на основе клинических испытаний in vitro с использованием стандартизированного вещественного фантома NEMA IEC с шестью сферами, имитирующими патологические очаги разных размеров. Однако возможности проведения таких многопараметрических экспериментальных измерений ограничены из-за высокойстоимости и лучевой нагрузки на исследователей. В данной работе развит альтернативный подход на основе имитационного компьютерного моделирования in silico с использованием цифрового двойника фантома NEMA IEC. Компьютерные эксперименты могут проводиться без ограничений с разными сценариями. По аналогии с клиническими испытаниями в численном моделировании оценивался коэффициент восстановления (RC_max), равный отношению максимального значения полученного решения в очаге к его точной величине. Условия моделирования были ориентированы на параметры клинических обследований методом ОФЭКТ/КТ с ^99mTc пациентов с заболеваниями и поражениями костной системы. Впервые выполнены исследования зависимости RC_max от величины отношения «очаг/фон» и влияния постфильтрации решения. В численных экспериментах были получены краевые артефакты на изображениях очагов, аналогичные тем, которые наблюдались при измерениях на реальном фантоме NEMA IEC и в клинической практике при обследовании пациентов. Краевые артефакты приводят к нестабильности поведения решения в итерационном процессе и к ошибкам в оценке накопления РФП в очагах. Показано, что постфильтрация снижает влияние этих артефактов, обеспечивая стабильное решение. Однако при этом существенно занижаются оценки решения в небольших очагах, поэтому предложено учитывать полученные в данной работе поправочные коэффициенты при количественной оценке активности в очагах диаметром менее 20 мм.

In single-photon emission computed tomography (SPECT), patients with bone disorders receive a radiopharmaceutical (RP) that accumulates selectively in pathological lesions. Accurate quantification of RP uptake plays a critical role in disease staging, prognosis, and the development of personalized treatment strategies. Traditionally, the accuracy of quantitative assessment is evaluated through in vitro clinical trials using the standardized physical NEMA IEC phantom, which contains six spheres simulating lesions of various sizes. However, such experiments are limited by high costs and radiation exposure to researchers. This study proposes an alternative in silico approach based on numerical simulation using a digital twin of the NEMA IEC phantom. The computational framework allows for extensive testing under varying conditions without physical constraints. Analogous to clinical protocols, we calculated the recovery coefficient (RC_max), defined as the ratio of the maximum activity in a lesion to its known true value. The simulation settings were tailored to clinical SPECT/CT protocols involving ^99mTc for patients with bone-related diseases. For the first time, we systematically analyzed the impact of lesion-to-background ratios and post-reconstruction filtering on RC_max values. Numerical experiments revealed the presence of edge artifacts in reconstructed lesion images, consistent with those observed in both real NEMA IEC phantom studies and patient scans. These artifacts introduce instability into the iterative reconstruction process and lead to errors in activity quantification. Our results demonstrate that post-filtering helps suppress edge artifacts and stabilizes the solution. However, it also significantly underestimates activity in small lesions. To address this issue, we introduce post-reconstruction correction factors derived from our simulations to improve the accuracy of quantification in lesions smaller than 20 mm in diameter.