MARKOV MODEL FORECASTING CONTROL OF TELECOMMUNICATION NETWORK SCALPING

Abstract

The article presents a Markov model of predictive control of the scaling of a telecommunication network, designed to describe the stochastic evolution of the network load and transitions between the main modes of the system's operation. The growth of traffic volumes and the dynamism of modern telecommunication systems necessitate the use of models that can take into account the random nature of changes in the intensity of data transmission and ensure timely decision-making on the expansion of network resources. In the proposed approach, the network operation is presented in the form of a discrete Markov process, where each state corresponds to a certain load level, and transitions between states are determined by a probability matrix that reflects the statistical characteristics of traffic.

The model allows taking into account various modes of network operation, including average load intervals, short-term peak surges and potential overload states. Based on k-step Markov transitions, a forecast of the future state of the system is formed, which makes it possible to estimate the probability of transition to critical modes and initiate scaling of computing or network resources in advance. This approach provides a transition from reactive to proactive control, which is important for maintaining stable quality of service in conditions of uneven and unpredictable traffic.

To verify the effectiveness of the proposed model, numerical simulations were conducted using input data close to the results of simulations in the OMNeT++ environment. The results obtained showed that the use of Markov forecasting allows reducing the system response time to peak loads by approximately 15% compared to traditional scaling methods based on current measurements of network parameters.

The results obtained confirm the effectiveness of the Markov approach for predictive control problems in telecommunication systems. The proposed model provides an adequate reflection of the dynamics of network traffic and can be used as an analytical basis for creating adaptive mechanisms for scaling resources in modern software-defined and cloud telecommunication networks.