Methods of Multidimensional Aggregation of Time Series of Streaming Data for Cyber-Physical System Monitoring

Data aggregation methods are developed and studied to increase the performance of intrusion detection systems in cyber-physical systems (CPS’s). What makes this work stick out is that it deals with the aggregation of data represented as time series with different periods in intrusion prediction and detection methods. The requirements on CPS data aggregation are given, and new methods of hierarchical and multidimensional streaming data aggregation are studied. Methods of multidimensional data aggregation based on trees and a directed graph are proposed and compared. For experimental evaluation, a prototype of a data aggregation system with systems of hierarchical and multidimensional aggregation is designed. The performance of the designed prototype is estimated, and the size of the required memory is given for each proposed method. An application technique is presented for the solutions with the parameters of the systems where the application of the developed methods is most efficiently described.