MODEL FOR ENSURING INFORMATION INTEGRITY IN IOT SYSTEMS BASED ON BLOCKCHAIN TECHNOLOGY USING A DISTRIBUTED REGISTRY TO REDUCE THE RISKS OF UNAUTHORIZED ACCESS

Abstract

Modern cyber-physical and information systems operate in conditions of dynamic parameter changes, uncertainty, incomplete input data, and significant noise influence, which complicates the process of forecasting and decision-making. Under such conditions, traditional prediction approaches often demonstrate insufficient adaptability and reduced estimation accuracy, especially in real-time systems. The paper proposes an integrated model for optimization and forecasting of dynamic processes based on the combination of a Kalman filter and a genetic optimizer. The proposed approach is aimed at improving prediction accuracy, adaptive parameter adjustment, and stability of decision-making under noisy and unstable operating conditions.

The developed model combines recursive state estimation mechanisms with evolutionary optimization methods. The Kalman filter is used for forecasting the system state and updating estimated parameters according to incoming measurements, while the genetic optimizer performs adaptive tuning of the model coefficients and minimizes the forecasting error. The architecture of the proposed model includes modules for data collection and preprocessing, state prediction, error estimation, adaptive optimization, and feedback correction. A fitness function based on minimizing the mean square forecasting error is applied to evaluate the effectiveness of optimized parameters.

An experimental study was conducted using several scenarios of dynamic process simulation with different noise levels and parameter instability conditions. The proposed model was compared with the classical Kalman filter and traditional optimization approaches. The obtained results demonstrated a significant reduction in forecasting error, increased resistance to noise disturbances, and improved adaptability to changes in system parameters. The integrated approach provided higher forecasting accuracy while maintaining acceptable computational complexity.

The practical significance of the developed model lies in the possibility of its application in decision support systems, embedded platforms, intelligent monitoring systems, sensor networks, autonomous control systems, and cyber-physical infrastructures. The scientific novelty of the research consists in the development of an adaptive integrated forecasting model that combines recursive state estimation and evolutionary optimization for dynamic parameter correction in real time.