Valery Mochalov

Dr. Valery Mochalov


Research Focus & Metrics

Academic Career Length

9Years (2018-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
24 March 2024 in Modeling of systems and processes
Verified
Journal Article
Closed access
VMValery Mochalov
Valery Mochalov
NBNatalia Bratchenko
Natalia Bratchenko
DGDar'YA Gosteva
Dar'YA Gosteva
A mathematical model and algorithm of a two-level load management system for virtual clusters of a data processing center (data center) have been developed. At the first management level, virtual machines (VMs) are assigned to physical servers. At the same time, a greedy algorithm is used with restrictions on the time of searching for acceptable load distribution alternatives. The second level of management is implemented taking into account the chaotic structure of network traffic between the data center and users. Checking for the randomness of a time series of information traffic is carried out using Lyapunov exponents. The predictive model of the load intensity is implemented using the method of phase space reconstruction based on a set of values of a one-dimensional time series. When constructing a reconstructed phase space attractor, the time delay value is selected from the condition of reaching the zero value of the autocorrelation function, and the dimension of the embedding is determined by the angle of inclination of the straight line approximating the dependence of the value of the correlation integral on the radius of a given threshold point. The Tayler window is used to exclude correlated points in the numerical series. The criterion for evaluating the effectiveness of the developed algorithm is an integral indicator of the deviation of the load of each server from a given level. The proposed model can be used to build a data center load balancing system in conditions of its nonlinear nature.

21 May 2018 in Journal of Physics: Conference Series
Verified
Conference Paper
Open access
PDF file
VMValery Mochalov
Valery Mochalov
NBN Y Bratchenko
N Y Bratchenko
SYS V Yakovlev
S V Yakovlev

4 June 2019 in Computers
Verified
Journal Article
Open access
PDF file
VMValery Mochalov
Valery Mochalov
NBNatalya Bratchenko
Natalya Bratchenko
GLGennady Linets
Gennady Linets
SYSergey Yakovlev
Sergey Yakovlev
The existing management systems for networks and communication services do not fully meet the demands of users in next-generation infocommunication services that are dictated by the business processes of companies. Open digital architecture (ODA) is able to dramatically simplify and automate main business processes using the logic of distributed computing and management, which allows implementing services on a set of network nodes. The performance of a distributed operational management system depends on the quality of solving several tasks as follows: the distribution of program components among processor modules; the prioritization of business processes with parallel execution; the elimination of dead states and interlocks during execution; and the reduction of system cost to integrate separate components of business processes. The program components can be distributed among processor modules by an iterative algorithm that calculates the frequency of resource conflicts; this algorithm yields a rational distribution in a finite number of iterations. The interlocks of parallel business processes can be eliminated using the classic file sharing example with two processes and also the methodology of colored Petri nets. The system cost of integration processes in a distributed management system is reduced through partitioning the network into segments with several controllers that interact with each other and manage the network in a coordinated way. This paper develops a model of a distributed operational management system for next-generation infocommunication networks that assesses the efficiency of operational activities for a communication company.

8 October 2025 in Russian Technological Journal
Verified
Journal Article
Open access
PDF file
VPV. I. Petrenko
V. I. Petrenko
FTF. B. Tebueva
F. B. Tebueva
MOM. G. Ogur
M. G. Ogur
GLG. I. Linets
G. I. Linets
VMValery Mochalov
Valery Mochalov
Objectives. The study sets out to develop a scalable method for detecting multi-vector attacks on Internet of Things (IoT) devices. Given the growth of security threats in IoT networks, such a solution must provide high accuracy in detecting attacks with minimal computing costs while taking into account the resource constraints of IoT devices.Methods. The developed hybrid neural network architecture combines convolutional networks for spatial dependence analysis and long short-term memory networks or gated recurrent units representing types of recurrent neural networks for analyzing time dependencies in network traffic. Model parameters and computational costs are reduced by pruning. A blockchain with a proof of voting consensus mechanism provides secure data management and decentralized verification.Results. Experiments on the CIC IoT Dataset 2023 showed the effectiveness of the model: the accuracy and F1 measure were 99.1%. This confirms the ability to detect known and new attacks in real time with high accuracy and completeness. Processing time is reduced to 12 ms, while memory usage is reduced to 180 MB, which makes the model suitable for devices with limited resources.Conclusions. The developed model is superior to analogues in terms of accuracy, processing time, and memory usage. Hybrid architecture, pruning, and decentralized verification provide effectiveness against multi-vector IoT threats.

21 April 2025 in Programmnaya Ingeneria
Verified
Journal Article
Closed access
VPV. I. Petrenko
V. I. Petrenko
FTF. B. Tebueva
F. B. Tebueva
MOM. G. Ogur
M. G. Ogur
GLG. I. Linets
G. I. Linets
VMValery Mochalov
Valery Mochalov
The paper presents a new scalable method for detecting multi-vector attacks on compromised Internet of Things (IoT) devices using machine learning algorithms. A detailed analysis of existing attack detection methods is conducted, including signature-based methods, anomaly-based methods, supervised and unsupervised machine learning methods, as well as deep learning methods such as convolutional and recurrent neural networks. A new method combining deep learning and federated learning is developed, which improves the accuracy, scalability, and security of the system in the context of limited IoT device resources. The results of an experimental study confirming the superiority of the proposed method over existing solutions are presented.

12 May 2026 in Advanced Engineering Research (Rostov-on-Don)
Verified
Journal Article
Open access
PDF file
GSG. V. Slyusarev
G. V. Slyusarev
VMValery Mochalov
Valery Mochalov
DGD. V. Gosteva
D. V. Gosteva
Introduction . Reliable identification of the probabilistic and temporal features of switching nodes is required for assessing the quantitative characteristics of software-defined networks. Widely used queuing theory (QT) methods only approximately specify and model the processes in an OpenFlow switch and its buffer memory. This results in understated and unrealistic performance estimates for the designed network equipment, causing switch buffer overloading and packet loss. A different modeling approach can solve this problem. The objective of this paper is to develop and study an analytical model for the buffer memory of an OpenFlow switch in an SDN using advanced techniques. Materials and Methods . The discrete Laplace-Stieltjes transform was used. Statistical characteristics of packet flows and the throughput of communication channels for a given packet loss probability were taken into account. The OpenFlow switch buffer memory model was based on the mathematical apparatus of the QT. It was constructed under the assumption of recurrence of input data flows with batch arrivals. The model was based on schematic representations of the switch structure, its record set, and a graph description of the transmission of network packets leaving the switch. We started with schematic representations of the switch structure, its record set, and a graph description of the transmission of network packets exiting the switch. Two model assumptions were taken as acceptable: arbitrary distribution of the relationship between the volume of data flows and their service time; discreteness of the distribution of the information flow structure. Results . The developed model integrated the probability of packet flow loss, their statistical characteristics, the throughput of computing devices, and the multiphase service procedure. When testing the model performance, we assumed that the switch load increased from 0.1 to 0.9, and the loss probability — from 10 –3 to 10 –6 . For these metrics, we determined how the switch load affected the buffer memory size and latency. In the first case, the minimum value (memory capacity) was 0.201, the maximum — 10564. In the second, they were 0 and 470 ms, respectively. For simulation modeling, the minimum time was 0 ms, the maximum — 2300 ms. The simulation and analytical modeling indicators were close at loads below 50% and increased several times at loads above 50%. The indicators increased sharply with loads up to 70%, and then increased exponentially. Discussion . At low network loads, queues did not overflow, packets were not lost, and linear dependences were maintained. At medium and high loads, packet flow processing was described by nonlinear dependences. The results of analytical and simulation modeling diverged due to the explosive nature of self-similar network traffic and its approximate description by the Pareto distribution. Switch load determined the feasibility of the proposed approach. The model is suitable for designing elements of software-defined networks to analyze their resilience under various information impacts. Conclusion . The proposed SDN analytical model determined the values and variances of the switch buffer memory size, as well as the memory capacity for constructing address flow tables. The solution performance was tested with switch loads ranging from 0.1 to 0.9. It is planned to create a model that takes into account request flows from both the external network and the server.