Defect prediction for embedded software

As ubiquitous computing becomes the reality of our lives, the demand for high quality embedded software in shortened intervals increases. In order to cope with this pressure, software developers seek new approaches to manage the development cycle: to finish on time, within budget and with no defects. Software defect prediction is one area that has to be focused to lower the cost of testing as well as to improve the quality of the end product. Defect prediction has been widely studied for software systems in general, however there are very few studies which specifically target embedded software. This paper examines defect prediction techniques from an embedded software point of view. We present the results of combining several machine learning techniques for defect prediction. We believe that the results of this study will guide us in finding better predictors and models for this purpose.