PLANNING THE EVOLUTION OF THE SOFTWARE DEVELOPMENT PROCESS BASED ON THE METHOD OF SEQUENTIAL OPTION ANALYSIS

Abstract

It is noted that the process approach is a means of scientific research and practical change of the surrounding world, which allows representing and implementing production systems in the form of a set of processes. The paper considers the peculiarities of using this approach to the study of human activity related to software development. A list of scientists, who drew attention to the necessity of implementing a process approach to software development, is presented, and on this basis, the concept of the software development process (SDP) was introduced. The concept of a model life cycle (MLC) is defined. It is emphasized that the most well-known sets of MLC are SMMI and SPICE maturity models. The improvement of SDP in the work is understood as a set of actions aimed at improving its characteristics by implementing a certain set of measures that are based on quality engineering, including TQM. It is emphasized that maturity models belong to one of two main categories: continuous and discrete (staged). The SMMI model is implemented on the basis of two approaches, and the SPICE model is implemented only on the basis of a continuous approach. Currently, the SMMI model has been formalized, which made it possible to plan the development of the SDP or its separate processes under conditions of limited resources. From the point of view of the SPICE reference model, the formalization of its structure is presented in the form of the following hierarchy: a set of process groups; multiple categories; set of processes; set of practices. Based on the given structure, the assessment of the possibility level of a separate process for SPICE model was formalized. The developed model became a basis for optimization problem development of planning evolution of a processes subset for the SPICE maturity model. In order to analyze this problem and develop an algorithm for its implementing, it is presented in a shortened form. Its objective function is additive and determines the integral indicator of the increase in the level of possibility of a processes subset over the entire planning period. The model has a resource limitation, which is characterized by the fact that resources that are not used in a given subperiod can be implemented in subsequent planning subperiods. It is shown that the method of sequential analysis of options, which can be implemented based on various algorithms, is the most suitable for the optimization of such a model. The work uses the "Kyiv Broom" algorithm, which is a multi-step process, at each planning sub-period of which "sweeping out" a certain set of non-competitive options is implemented.