A Decentralized Evolutional Approach to Handle Schedule Execution in Software Projects

摘要

We present an application of multi-agent to the problem of software project schedule execution evolution, i.e. performing an automatic schedule monitoring and handling contingencies that result in project delivered on time, and required specifications. In a software project setting each involved individual can be represented by three cooperative agents: a proxy, a bookkeeper and a scheduler. These cooperative agents adopt a proactive schedule representation that is able to withstand task execution delays, and use reactive cooperative adaptation strategies with other teammates having bonded tasks activity to generate a schedule solution that maximizes the team reward. Our experimental results with real software project instances data underline the validity of our approach.