Abstract
The authors propose a stepwise statistical methodology for the study of operating system reliability and associated tools. An example of the application of this method for the ARGOS data processing center of France's CNES is presented. It is shown that each evaluation of software reliability is considered as a special case. Two major consequences are that the reliability models need improvement and that every evaluation must be supported by a statistical analysis of the product and process characteristics.

This publication has 0 references indexed in Scilit: