Agent Negotiation Techniques for Improving Quality-Attribute Architectural Tradeoffs

ARIEL MONTESERIN; SILVIA SCHIAFFINO; IGNACIO GATTI; J. ANDRES DIAZ PACE

Porto

Conferencia; 15th International Conference on Practical Applications of Agents and Multi-Agent Systems, PAAMS 2017; 2017

A key aspect of software architecture design is to satisfy quality-attribute requirements, such as performance or modifiability. This is usually a complex task, because there are often several candidate solutions meeting the same requirements, and the quality-attribute tradeoffs of those solutions need to be considered by architects. In previous work, we developed an agent-based approach called DesignBots to assist architects in the exploration of design solutions driven by quality attributes. The agents performed a local search, each one optimizing a particular quality, but they were limited regarding tradeoff analysis capabilities. In this paper, we propose negotiation techniques for improving the tradeoff analysis of the agents, inspired by how human architects (and stakeholders) normally work in real-life projects when assessing alternative designs. In particular, we develop two negotiation strategies that integrate with the DesignBots framework. The experimental results obtained so far with an architectural case-study show that the negotiation can produce more satisfying tradeoffs than those currently provided by DesignBots.