A MARKOVIAN SOFTWARE AVAILABILITY MEASUREMENT WITH A GEOMETRICALLY DECREASING FAILURE-OCCURRENCE RATE

We develop a software availability model incorporating software failur e-occurrence and fault-correction times, under the assumption that the hazard rate for software failure-occurrence decreases geometrically w ith the progress in fault-removal process. Considering that the softwa re system alternates two states, i.e. the operational state that a sys tem is operating and the maintenance state that a system is inoperable due to the Fault-correction activity, we model the time-dependent beh avior of the system with a Markov process. Expressions for several qua ntities of software system performance are derived from this model. Fi nally, numerical examples are presented for illustration of software a vailability measurement.