Summary: The article considers a fragment of a computer network whose operation is described by a stochastic process with a finite number of states. A method is proposed for determining the probabilistic moments of the accumulated residence time of the fragment in a fixed subset of states, and the probabilistic moments of the number of times the fragment enters the states of this subset.
The method is based on the results of analysis of linear objects with a variable mode of operation. Cases of inhomogeneous Markov and semi-Markov processes of change of states are examined from a unified standpoint. An example of the application of the results to the analysis of a subscriber installation is given.