Asideh F. Models and methods of increasing dependability informational-control systems of critical application by use of multichannel duplication

Українська версія

Thesis for the degree of Candidate of Sciences (CSc)

State registration number

0405U001062

Applicant for

Specialization

  • 05.13.06 - Інформаційні технології

25-02-2005

Specialized Academic Board

Д64.062.01

Essay

The dissertation is devoted to development of dependability assessment models, methods of redundancy and restoration of information-control systems with multichannel structural duplication (ICS-MSD) taking into account hardware (HW) and software (SW) faults. Scientific results are: 1) model of ICS-MSD, determining system connections of functional and dependability attributes of fault-tolerant architectures and allowing to synthesize their systematized set taking into account number and distributions of SW versions on channels, completeness of duplication, ability to adaptation; 2) method of adaptation of ICS-MSD to HW and SW faults, basing on dynamic reconfiguration and procedure of consecutive fixing and the analysis of failure syndromes and providing preservation of work capacity at failures of any of two channels; 3) availability markov's models of restored ICS-MSD, taking into account strategy restoration parameters, variants of introduction of version redundancy and the subsystem check organization; 4) set of ICS-MSD restoration strategies taking into account the moments of the beginning, volume and priority of restored HW and SW faults; 5) method of ICS-MSD architectures and restoration strategies choice supporting search of rational structural and operational parameters on criteria of availability and fault tolerance. Offered methods and means allow to increase fault tolerance and availability of ICS of critical application at the expense of complex use of procedures of adaptation, multiversion technologies and flexible restoration strategies on HW and SW faults and to automate decision-making processes at designing such systems.

Files

Similar theses