Résumé
Real-time computations in multi-agent systems have been studied from different perspectives of reasoning, message passing , resource management and negotiations. Separate treatment of all these perspectives may result in performance amelioration in one dimension while over-passing others. Here we propose an integrated “real-time” mechanism where all dimensions of temporal behavior are suggested in one unified model. Here in this position paper, we argue for an integrated and comprehensive treatment of “real-time” issue under the title of coordination. Our position regarding real-time agent coordination would result in overall performance betterment and overhead lessening, if all sub-processes are made to respect “notion of time” in design. We analyze current approaches and options as well as present an outline for an integrated and comprehensive view of “real-time coordination”.