A Model of Distributed Disorders Detection
share
The paper deals with disorders detection in the multivariate stochastic process. We consider the multidimensional Poisson process or the multivariate renewal process. This class of processes can be used as a description of the distributed detection system. The multivariate renewal process can be seen as the sequence of random vectors, where parts of its coordinates are holding times, others are the size of jumps and the index of stream, at which the new event appears. It is assumed that at each stream two kinds of changes are possible: in the holding time or in the size of jumps distribution. The various specific mutual relations between the change points are possible. The aim of the research is to derive the detectors which realize the optimal value of the specified criterion. The change point moment estimates have been obtained in some cases. The difficulties have appeared for the dependent streams with unspecified order of change points. The presented results suggest further research on the construction of detectors in the general model.
READ FULL TEXT