Comparing Agent Software Development Methodologies using the CMMI Engineering Process Model

This paper explores the fulfillment of Capability Maturity Model Integration (CMMI) processes and practices by two widely published agent-based software development methodologies: Multiagent Systems Engineering Methodology (MaSE) and Gaia. Comparisons of these methodologies have been documented and published, but no known research exists that compares and contrasts these methodologies with the CMMI processes. CMMI provides a benchmark for mature, capable processes, identifying practices necessary to implement those processes effectively. In this paper, the engineering process areas of CMMI are used as a baseline for a theoretical analysis of capabilities offered by software development methodologies designed specifically for the agent-based software domain. An overview of the engineering process areas along with general descriptions of MaSE and Gaia are included. Differences between these methodologies are explored in a functional mapping to the engineering process areas of CMMI. The analysis will show that, while each methodology examined offers guidance in the design and development of solutions, guidance for implementation is minimal. Additionally, verification and validation techniques, critical to effective software engineering, are not addressed sufficiently by MaSE or Gaia.