Application of kalman filter in time series software reliability growth model

Researches show that assumptions condition of existing software reliability growth models are difficult to be satisfied in actual projects which restrict the universality of models. Classical models neglect observation noise and its affection on accurate evaluation to software reliability. This paper proposes a time series software reliability growth model and transforms it into state space model and Kalman filter is used to reduce noise. Testing data of filtering noise can shows the essential rule of data better and improves goodness of fit. Simulation result shows the validity of this method.