2020
Permanent URI for this community
Browse
Browsing 2020 by Author "Pakhomova, Victoria M."
Now showing 1 - 2 of 2
Results Per Page
Sort Options
Item Detection of Attacks on a Computer Network Based on the Use of Neural Networks Complex(Дніпровський національний університет залізничного транспорту імені академіка В. Лазаряна, Дніпро, 2020) Zhukovyts’kyy, Igor V.; Pakhomova, Victoria M.; Ostapets, Denis O.; Tsyhanok, O. I.ENG: Purpose. The article is aimed at the development of a methodology for detecting attacks on a computer network. To achieve this goal the following tasks were solved: to develop a methodology for detecting attacks on a computer network based on an ensemble of neural networks using normalized data from the open KDD Cup 99 database; when performing machine training to identify the optimal parameters of the neural network which will provide a sufficiently high level of reliability of detection of intrusions into the computer network. Methodology. As an architectural solution of the attack detection module, a two-level network system is proposed, based on an ensemble of five neural networks of the multilayer perceptron type. The first neural network to determine the category of attack class (DoS, R2L, U2R, Probe) or the fact that there was no attack; other neural networks – to detect the type of attack, if any (each of these four neural networks corresponds to one class of attack and is able to identify types that belong only to this class). Findings. The created software model was used to study the parameters of the neural network configuration 41–1–132–5, which determines the category of the attack class on the computer network. It is determined that the optimal training speed is 0.001. The ADAM algorithm proved to be the best for optimization. The ReLU function is the most suitable activation function for the hidden layer, and the hyperbolic tangent function – for the output layer activation function. Accuracy in test and validation samples was 92.86 % and 91.03 %, respectively. Originality. The developed software model, which uses the Python 3.5 programming lan-guage, the integrated development environment PyCharm 2016.3 and the Tensorflow 1.2 framework, makes it pos-sible to detect all types of attacks of DoS, U2R, R2L, Probe classes. Practical value. Graphical dependencies of accuracy of neural networks at various parameters are received: speed of training; activation function; optimization algorithm. The optimal parameters of neural networks have been determined, which will ensure a sufficiently high level of reliability of intrusion detection into a computer network.Item Organizing Wireless Network at Marshalling Yards Using the Bee Method(Dnipro National University of Railway Transport named after Academician V. Lazaryan, Dnipro, 2020) Pakhomova, Victoria M.; Nazarova, Diana I.ENG: Purpose. In general, today wireless networks are widely used as an alternative to wired, allowing you to connect multiple devices, both among themselves in the local and global Internet. However, at the present stage in Ukraine there is no widespread use of a wireless network at rail transport, therefore it is advisable to conduct research on the deployment of such a network, in particular, at a marshalling yard. Methodology. Using LocBS-BeeCol program model written in Python according to the bee colony algorithm the optimal number of base stations (BS) of the wireless network and their location at the marshalling yards was determined, as well as research on the bee algorithm parameters was conducted. Input data of the LocBS-BeeCol model are as follows: marshalling yard parameters (area, number of clients that need to be connected to base stations); wireless network parameters (base station coverage radius, maximum number of clients for one base station); parameters of the bee colony algorithm (number of scout bees, number of attempts to find the optimal solution using one bee). Findings. For marshalling yards of various capacities (small, medium and high), the optimal number of base stations of the wireless network was obtained with restrictions on the coverage radius of the base station and the number of clients connected to it. Thus, for example, to connect 300 clients at medium-sized marshalling yards with an area of 2500x500 m2, 93 base stations with a coverage radius of 50 m are needed. Originality. The quality of the obtained solutions significantly depends on the choice of the bee colony algorithm parameters. A study of the base stations number of the wireless network and search time for finding the optimal solution for different number of bees and the number of attempts to find the op-timal solution using the bee for marshalling yards of various capacities was carried out. It was determined that an increase in the number of bees (from 10 to 50) and the number of attempts to find the optimal solution by a bee (from 10 to 50) improves the quality of the optimal solution (decrease in the number of base stations by an average of 6.5% and 9.3%), respectively. In addition, increase in the bee number (from 10 to 50) reduces the search time for the optimal solution by bees by an average of 1.8 times, while increase in the number of attempts to find the optimal solution by a bee (from 10 to 50) will increase search time for the optimal solution on average 2.14 times. Practical value. An algorithm and its software implementation have been developed, which make it possible to determine the required number of base stations and their location when deploying a wireless network at a marshalling yards. For marshalling yards with high capacity, when the coverage radius of the base station is doubled (from 50 to 100 m), their number decreases by about half (from 136 to 64), while the time for finding the optimal solution by bees increases by 2.5 times (from 8.4 to 20.6 s).