Modeling of SW reliability in early design with planning and measurement of its reliability growth

Abstract

This paper presents a method for early software reliability assessment with planning and measurement of achieved reliability improvement during the development process of the software. Early assessment models for software (SW) reliability improvement contain parameters which are estimated using the existing records of reliability improvements, as related to the software size and applied SW development processes for its fault mitigation and management. SW reliability is then projected through its maturity when its reliability reaches an approximate steady state where further reliability growth would be too small to justify further improvements. Reliability growth planning model, which was originally developed for planning and monitoring of reliability growth of hardware systems in their design phase, uses the same mathematical approach as the SW reliability estimate [3]. This model is modified for planning of SW reliability growth. Data on reliability improvements from incremental SW testing during development and the information about SW reliability improvement resultant from SW Failure Modes and Effects Analysis (FMEA) actions monitor and measure achieved reliability growth. Reliability growth measurements are then used to refine assumptions used for determination of the SW reliability model parameters. The SW reliability model is compatible with the hardware reliability growth mathematical model so that the estimate of the system level reliability growth (HW and SW) is easily achievable.