Stokipny A. Diagnosis of corporate IP-network services using regularities in components functioning

The thesis work deals with elaboration of a diagnostic model and methods to identify regularities in corporate IP-network function components for raising its availability by means of time reduction when searching for services' defective condition reasons. The work gives further development to the corporate IP-network diagnostic model able to describe (as opposed to the existent models) the condition of the corporate IP-network and its services as a set of events. This makes it possible to reduce the number of considered services' defective conditions by dealing with those essential for the diagnosis process. The developed method of identification of regularities in the corporate IP-network components functioning allows to get additional information about the defects and their causes. Further to this development, we suggest and implement a method to search for the sets of events allowing to identify the causes of the corporate IP-network services' defective conditions at a specified probability (with the specified reliability). An improved the method to search for frequent closed item sets in the content rule search tasks. A method to construct a decision support system facilitating the corporate IP-network maintenance was elaborated. We offer recommendations to implement the suggested diagnostic model and methods in the DSS.