Model-Based Task Allocation in Distributed Software Development

Task allocation is one central aspect in planning and managing global software development projects. To date several models that support task allocation have been proposed, including cost models and risk-based approaches. However, systematic integration of such models and a guiding process for task allocation activities is widely missing. In this article, we integrate existing models that reflect different viewpoints and abstraction levels of task allocation decisions. Based on the integrated approach, we sketch a process for systematic evaluation and selection of task assignments that defines the model interfaces and the sequential order of their use. In detail, the approach presented here integrates a risk model that is able to identify the possible risks for each assignment individually, an optimization model that uses Bayesian networks to suggest assignment alternatives with respect to multiple criteria, and an effort overhead model that is able to estimate the project effort for each assignment alternative. All three models are significantly grounded in empirical studies. Besides the introduction of all three models and the description of the process, the article provides an example application, sketches related work, and presents an overview of future work.