Кафедра "Електронні обчислювальнi машини" (КЕОМ ДІІТ)

Permanent URI for this communityhttp://crust.ust.edu.ua/handle/123456789/695

ENG: Department "Electronic Computers"

Browse

Now showing 1 - 5 of 5

Determination of Network Attacks Using Neural Network Technologies
(ScientificWorld-NetAkhatAV, Karlsruhe, Germany, 2021) Pakhomova, Victoria N.
ENG: Formulation of the problem. Intrusion-Detection Systems (IDS) are used to detect network attacks in real time. In the information and telecommunication system (ITS) of railway transport, the problem of a large volume of network traffic arises, since standard approaches to data processing cease to be effective. One of the most effective approaches to classifying a large amount of data is the use of neural network technology. This approach allows detecting not only already known network attacks, but also detecting new ones.
Distribution of Information Flows in the Advanced Network of MPLS of Railway Transport by Means of a Neural Model
(Dnipro National University of Railway Transport named after Academician V. Lazaryan, 2019) Zhukovyts’kyy, Ihor; Pakhomova, Victoria N.; Domanskay, Halyna; Nechaiev, Andrew
ENG: Abstract. Ensuring interoperability of railway transport is possible only due to the developed information structure. Today, Ukraine uses the information-telecommunication system (ITS) of railway transport, which is based on a data communication network. The effectiveness of its work is largely determined by the routing system. The current algorithm for choosing the shortest route, which is used in the existing routing protocol (OSPF), does not always lead to an effective result. However, there is MPLS technology, which could improve the quality of the ITS network by creating virtual channels between its nodes. The authors proposed a scheme for selecting tunnels for the flows in the MPLS network, which is based on the neural model of a multilayer perceptron of configuration 18–3–3–10 with the activation function Softmax in hidden layers and a linear activation function in the input layer. To simulate the network operation, flow data is needed: class of service (CoS), sender and recipient identifiers, average flow rate vector and tunnel data (their initial load). The final load of the tunnels is taken as the resulting output of the neural network, on the basis of which the tunnel is selected for the flow of the k-th class of service.
Research of Token Ring Network Options in Automation System of Marshalling Yard
(Politechnika Slaska, Poland, 2018) Zhukovyts’kyy, Igor V.; Pakhomova, Victoria N.
ENG: In the automation systems of sorting process the use of computer networks of Ethernet technology is possible. But its traditional drawback is a probabilistic nature of the network access, which does not guarantee the information transmission in specified time intervals. However, there are other technologies that are free from this defect, for example, the Token Ring technology. A methodology, for determination of the network parameters in the automation system on marshalling yard at the restriction of the average time for application waiting in the queue at the station network has worked out. Using the token method of access to the ring, the state diagram of network stations was compiled. This is the basis for simulation model of the Token Ring network, which was developed in the GPSS environment. As a result of simulation the dependencies of the maximum queue length at the network station on the frame length, intensity, and distribution of time for application receipt from the station were obtained.
Research on the Possibility of the Bee Colony Algorithm for Determining the Topology of the Wireless Network at the Marshalling Yard
(Faculty of Management Science and Informatics, University of Zilina, Slovakia, 2020) Nazarova, Diana I.; Pakhomova, Victoria N.
ENG: For railway marshalling yards of different power (low, medium, high), an optimal number of wireless base stations and their location were determined on a Python program based on a bee colony algorithm. Program input: marshalling yard parameters (area, number of clients); wireless network parameters (coverage radius and number of base station clients); parameters of the algorithm (number of bees, number of attempts). For example, to connect 300 clients at the medium-power marshalling yard, 93 base stations with a coverage radius of 50 m are required. The quality of solutions depends heavily on the choice of parameters of the bee colony algorithm. It is determined that increasing the number of bees (from 10 to 50) and the number of attempts to find the optimal bee solution (from 10 to 50) leads to an improvement in the quality of the optimal solution (reducing the number of base stations by an average of 6.5% and 9.3%, respectively). In addition, increasing the number of bees by 5 times leads to a decrease in the search time of the bee optimal solution by an average of 1.8 times, while increasing the number of attempts to find the optimal bee solution by 5 times will increase the search time of the solution by an average of 2.14 times. In particular, for the high-power marshalling yard, when the base stations coverage radius is doubled (from 50 to 100 m), their number decreases approximately twice (from 136 to 64), while the search time for the bee optimal solution is increased by 2.5 times (from 8.4 to 20.6 s).
Study of the Possibility of Using the RBF Network to Detect U2R Category Network Attacks
(D.A. Tsenov Academy of Economics, Svishtov, Bulgaria, 2022) Pakhomova, Victoria N.; Kulyk, Victoria
ENG: The "RBF_U2R" program based on the implementation of the RBF network, the configuration of which is N-M-K (where N is the number of input neurons; M is the number of basic functions; K is the number of resulting neurons) was created in Python for detecting the following classes of attacks: Buffer_overflow; Loadmodule; Perl; Rootkit; Normal and using network traffic parameters from the open KDDCup database. Studies of the accuracy parameter were carried out during the training epochs of the neural network on the created program.