Agena Company Logo

Improved monitoring and prediction of software defects

Problem

• Software is increasingly used to build embedded components which may be safety-critical (such as in transport systems) or business critical (such as in electronic devices like TVs and DVD players)
• These embedded software components are getting larger and more complex
• Companies like Philips must be sure that defects in embedded software components are kept to an absolute minimum, since recall of such devices is impossible
• Organisations building safety critical components must be able to satisfy regulators than the number of defects is minimal
• In all such case software project managers need to know, with confidence, how much more testing is required before the software can be released
• They also need to know, with confidence, how reliable the software will be in operation

Solution

• We developed a class of risk maps in AgenaRisk that used information available at all stages of the software life cycle to monitor and predict defect risks
• AgenaRisk modelled the processes of defect insertion and discovery at the software module level
• AgenaRisk predicted the number of residual defects at various life cycle phases and with various different types of assumptions about the design and testing process

Benefits

• At Philips AgenaRisk's predictions on a range of projects were 95% accurate. Specifically the correlation between defects predicted by AgenaRisk and actual defects discovered was 95%. Other approaches achieve at best 70% accuracy
• AgenaRisk supports process improvement assesments and decisions - accurate prediction means that testing and rework effort can be assigned more efficiently
• Overall effect of more accurate predictions is higher software quality and lower testing costs

• © 2007 Agena |
• Privacy Policy |
• About Us |
• Contact Us