ON GENERATING VARIABLE-STRUCTURE ARCHITECTURES FOR DECISION-MAKING SYSTEMS

A quantitative approach for modeling a class of variable structure dec ision-making systems and for generating algorithmically alternative ar chitectures is presented. These systems, whether they are human organi zations or decision-making and information processing systems, consist of inter-connections of intelligent nodes; the interactions between t he nodes can change depending on the task being processed. Colored Pet ri Nets are used as the mathematical framework for representing design requirements and for modeling the variable structures. An algorithm i s presented for generating the complete set of variable structures tha t satisfy both some generic structural constraints as well as problem- specific design requirements. The model and the algorithm yield some i nsights on the properties of these systems and the information require d for their coordination for distributed decision-making. An applicati on to a non-trivial example for an Airport Surface Traffic Control Sys tem is presented.